Blog - Combine | Combine


Courtesy of Tesla, Inc.



Brains – Bold decisions in the modelling phase

A major redesign of their machine learning models includes a multi camera neural network making predictions directly in vector space. Previous models made predictions in image space for each camera, and the results were then transformed to vector space in post-processing.

This holistic approach allows the model to simultaneously make predictions on views from all cameras, similar to how the human brain uses vision from both eyes when making decisions.

Figur 1: It’s basically night and day, you can actually drive on this” says Andrej while also highlighting the huge engineering effort required to get this right.

Tesla also explains how a lack of memory has been a problem for their previous AI models, citing issues such as temporary occlusion. By redesigning their models to incorporate so called

hidden states capable of memorizing information from previous frames in the video the accuracy improved.

Figur 2: Using the context in video. Andrej Karpathy demonstrating how using the context in video reduces false detections due to occlusions.

Muscles - Huge investments in data collection, labeling infrastructure and digital twins

After working with third party labelers Tesla now favors hiring professional labelers who are co-located with machine learning engineers. This allows Tesla to further develop and improve annotation software, as showcased in their 3D annotation tool.

Figur 3: Tesla’s 3D annotation tool capable of simultaneously annotating multiple camera views by making use of data captured from different cars.

Data is collected from the huge participating fleet of cars in real world traffic. Data can be collected on demand based on queries specified by the engineers, allowing specific hard cases to be improved within weeks rather than months. Offline analytics allows Tesla to automatically label much of their data, while still allowing human intervention where necessary.

For traffic scenarios not easily captured from their fleet Tesla uses simulation software to create realistic scenarios. These simulations realistically model real sensor noise, motion blur and other optical distortions.

Figur 4 & 5: Realistic simulation allows Tesla to render scenarios with perfect labels – saving annotation efforts. In this simulated intersection the not yet released Tesla Cybertruck can be spotted.

Speed – Computing power on the edge and in the Dojo

Making use of these efforts require a huge amount of compute power, something Tesla with their Project Dojo has developed their own scalable, distributed, processing architecture for. The supercomputer Dojo will feature their in-house designed D1 chip. More info can be found here:

Inspired? We can help you with your next step!

Whether you are beginning your AI journey or already have models in production, you can turn to Combine for expert help. Feel free to check out our blog posts about how to use virtual modeling for AI:

Large scale annotation using crowd-sourced data annotation and image classification using deep learning:


Want to learn more about how Combine can assist in AI development? Contact Simon Yngve (

Read more

I have always been interested in technical gadgets, even before my commodore 64 I took things apart to see how they were built up. In the earlier years I rarely put the gadgets back together. The C64 survived though, just had to adjust the cassette deck every now and then. I guess that can give you a hint regarding how old I am.

My career has revolved around software, from user requirements to system verification. Always in or related to embedded devices. My time at Baxter taught me a thing or two about software for safety critical systems. In that case for medical devices but it seems similar to the standards for the automotive industry.

I took the job at Combine Control Systems because I think they have a unique setup with their competencies within control systems and data science. I also liked that I get to build my own team of embedded developers from scratch. The competencies in my team will hopefully complement Combine’s current strengths. This will prepare us to take on even more fun and challenging projects.

Transparency and empathy are important to me. Be direct but on the other person’s wavelength. Address problems but focus on the solution rather than lingering on the past.

I’m also striving for making the teams I’m involved in a little more fun and interesting.


Feel free to reach out to me regarding any prospects, whether you want to join my team, want us to take on a whole project or want to hire myself directly for a role.

Read more

What I like most about Combine is that they take care of their employees and listen to them to create a great workplace. They are also very clear about which competences they possess, so that both new employees and clients know what to expect. My job will focus on creating long lasting relationships with all parties. The employees shall know that their assignments will be chosen from an individual view, with their background and experiences. Clients shall know what Combine stands for and what we contribute, which is more of partnership than a customer/supplier relationship.

But then again, we are suffering from Covid-19, which is affecting all industries and companies, not least in the consulting sector. How I am supposed to develop a company under these circumstances? One should be humble when facing this pandemic, but my strategy will be to make the new become the normal. Combine as a company is used to digital meetings and projects performed from a distance, for example doing work for American and Asian companies. We have the potential to manage projects like these even on a local scale, which renders even better prerequisites for satisfied clients.

I will put Combine on the map, at least on a regional scale. I look forward to getting to know all the contacts I will make in my new position. Before the end of 2021 we shall double the amount of people tied to our office in Linköping and during the upcoming years we look forward to doubling the workforce again. I am excited that, together with Rikard Hagman, carry through our vision for Combine and I believe that there are people who feels just like me. Let us know so we can do this together instead of waiting for Combine to headhunt you.

The upcoming year will be one of my most thrilling years. I believe that many people will move between companies and we will work proactive to connect with the people we want to grow with. I want to represent good leadership and mediate a great feeling, a transparent way of working and supply possibilities to develop.

Read more

Large industrial plants consist of hundreds if not thousands of sensors, constantly collecting data from the numerous processes that make up the plant. These sensors then transmit this data not only to the control room, but also to controllers placed throughout the plant. In turn, these controllers use this information to calculate the appropriate control signals and send them to the numerous actuators of the plants. These actuators, consisting of pumps, heaters, valves, etc., then apply the control signals on the industrial processes. Hence, it can be said that there is a mass of communication happening during the running of an industrial plant, as signals are consistently sent from sensors to controllers and from controllers to actuators.

These communication requirements pose challenges when designing the plant. Traditionally the communication has been done using wires to connect the sensors, controllers, and actuators where appropriate. However, this is an expensive and space-intensive solution. Furthermore, wired communications are inflexible as if one wishes to move a single sensor to a different location, this may necessitate new wiring. Therefore, there are unquestionably considerable benefits to be gleaned from utilizing wireless communications for industrial plants.

However, wireless communications pose limitations on their own. The chief among them is the question of reliability. Wireless communications are generally considered less reliable than wired communication. This means that with wireless communication, the communication channels between the controller and actuator and between the sensor and controller may be subject to the possibility of packet losses and packet delays.

With wireless communication, there may be an unreliable channel between the sensors and the controllers and between the controllers and the actuators.

Control in the case of unreliable channels poses a problem in itself.  The unreliability of the channel forces us to consider how to adapt our control scheme to compensate for the fact that signals sent by the controller may arrive to the actuator delayed or may not arrive at all. Furthermore, the delays from unreliable communication channels tend to be stochastic, so conventional control strategies for handling constant delays cannot be used.

Is LQG control the solution?

LQG control is an attractive proposition as it is based on the expected outcome, and hence can be expanded to consider the stochastic nature of the communication channel. Furthermore, if one assumes that acknowledgments ensure that the controller knows which signals have reached the actuators, the separation principle can be proved to hold. This is a key principle of LQG control, which states that the control problem and the state estimation problem can be solved separately. What this means in effect is that we do not need to concern ourselves with the unreliability of the communication channel between the sensors and controllers when deriving our control input. Conversely, we do not need to concern ourselves with the unreliability of the channel between controller and actuator when deriving our state estimates.

For state estimation, only relatively minor modifications to the conventional Kalman filter is required to compensate for the possibility of packet delays and packet losses between the sensor and controller. However, when implementing LQG controllers, the required modifications are somewhat more extensive. In this case, the optimal control signal depends not only on the states (as in conventional LQG control) but also on past inputs sent by the controller but have not yet reached the actuator.

If we evaluate our optimal LQG control solution (modified to take delays and package losses into account) and compare it to the standard LQG solution on a model of an inverted pendulum, we get the result below. As can be seen from the figure below, compensating for the delays results in considerably less deviation from the reference while requiring far less control input. Thus, the importance of taking the unreliability of the channel into account can be shown.

Comparing the optimal LQG solution (adapted to take into account the unreliability of the channel) with the standard LQG control.

Read more


As the developers have published a version in PyPI, installing the Accelerator now is as simple as running pip! As we are mostly using Python3 and we are testing and developing most of our code locally before sending it to our processing server, we can simply install it like this:

pip3 install –user accelerator


Then, the accelerator tool becomes available with the `ax` command!

First steps

Once the accelerator is intalled, it’s time to configure it! The new version if fully documented in the manual, but as a starting point it should be enough to run `ax –help` and do the same for the small subset of commands provided. Then, to actually get started, we run `ax init` in the working directory where we wish to create our first project. Usually, we would configure the accelerator locally for every project, and therefore run all the comands from the corresponding directory. As that is not necessarily the case for every developer, one could also setup a global configuration file (e.g: /etc/accelerator.conf) and simply add `alias ax=ax –config /ect/accelerator.conf` to the .bashrc file (assuming bash as the default shell).

Running `ax init` will create a set workdirs and files in the directory where it is run. The most important there is `accelerator.conf`, which contains the basic configuration. There we can configure thinks like the amount of threads that we want to run or where we want to the results of our scripts to land. Luckily, the stardard one created by `ax init` contains sane defaults and comments that help understanding what the different options do.

Finally, to start the server process, it is enough to run `ax server` from the project directory. Now the server is listening and to send some work, run `ax run “PIPELINE”` from another terminal. Starting running the tests can be a good start: `ax run tests`.

Further improvements

The new changes to the accelerator do not only come from the installation side, but also from the User Interface part. Now, it is possible to get detailed information about the tasks that have been run, their options, datasets, or results from the browser. For that, it is as simple as running `ax board` in the corresponding working directory or setting the `board listen` parameter in accelerator.conf. The board will start a server process that is by default listening on localhost:8520, which is of course customizable. There, we can check the status of the accelerator server process and search the results that were returned by our jobs, the options that were passed or which dataset was used. In addition, it’s even also possible to see the source code of the script that was imported to run a specific tasks. All this together, allows for a much better user experience and greatly helps debugging, specially when getting started using the tool. 

Some additional extra improvements include adding a `grep-like` command to search for patterns in datasets and merging urd’s (the part that keeps track of tasks) functionality into a single tool, not requiring urd to be set up independently anymore or the ability to configure the python interpreter for every method.

Read more

To understand this, we need to define a broader term ‘Vehicle Energy Efficiency’. Fuel Efficiency is generally described as a ratio of distance traveled per unit of fuel consumed (source: Wikipedia). Since fuel is a form of energy anyway, this term applies to fuel efficiency as well, albeit in a broader sense. When we consider the vehicle’s energy efficiency as a whole, other factors such as carbon dioxide emissions, output energy over a drive cycle, amongst others, come into play. Let’s face it, owning a car is a luxury and a privilege, irrespective of the number of arguments we stack up to defend such a statement. “I need the car to pick up and drop off the kids every day”, “Without the car, my wife and I would find it increasingly difficult to manage both our busy schedules”, some would say. Increasingly demanding, yes! But not impossible. Because here’s the thing with cars: THEY LEAVE SOME SERIOUS CARBON FOOTPRINT BEHIND!

According to the European Union official website, passenger cars account for around 12% of the total EU emission of carbon dioxide. There is thus, a dire need for more stringent regulations for the emissions of passenger cars. This primarily motivates the auto industry to focus more on “energy efficiency” to meet the target requirements of each new year. Having worked on an assignment at Volvo Cars within the Vehicle Energy Efficiency team helped me gain a deeper understanding of how the reduction of emissions and setting stricter targets every year is crucial in the long run to contribute positively toward a cleaner environment.

A typical workday would go on like this: ensure that the correct vehicle models were received to be simulated and to verify if the road load coefficients matched the models received. Once all the vehicle models with all the necessary parameters were obtained, then came the main and the most crucial step: running the simulations to obtain results for Fuel Consumption, CO2 emissions, and other relevant parameters which give us an idea of how much emissions each vehicle model belonging to a fleet or market was releasing. Sometimes, a part of the job description also included something called “Correlation Analysis”. The main goal of such an analysis was to “correlate” the results from the virtual tests we’d performed (via simulations) to the results of the physical tests carried out in the testing lab. This was primarily done using Sympathy For Data, Combine’s internal tool for data processing, and data visualization.

Anyway, the results of the simulations run, as well as the correlation analysis, were then presented to the Attributes Realization Engineers, who would then draw a comparison of the models to see which did and which did not meet the emission certification standards required. Keeping the fierce spirit of competition in mind, it was thus a no-brainer that the emission numbers of other auto manufacturers were compared. What possible tuning of parameters could lead to better numbers was discussed.

The goal of reducing emissions is of such a high priority to these automakers that there is even a comparison of emission numbers for vehicles of the same model but having a small difference in components for eg., in my case, it was generally a comparison between a vehicle having a 7-speed gearbox and a similar vehicle having an 8-speed one. While this change might seem insignificant and small, a different gearbox introduces a change in the vehicle mass, the gear ratios become entirely different, and of course, the gear shifting RPM changes as well, all thus directly or indirectly contributing to minute change in fuel economy per drive cycle or road trip.

Thus, the underlying goal remained the same: if there was any way, any way at all the emissions of the vehicles could be possibly reduced, go for it! Of course, the actual numbers I cannot mention here as such information is generally considered “hush-hush”. But I’m assuming you get it: Cars are something which people will only keep buying and buying either for their convenience or even indulgence. And because vehicles, in general, have become so embedded into the web that we like to call life, to think that we will, on a fine Saturday morning, reduce or possibly even stop the usage of passenger cars altogether is a farce. Even the new Electric Vehicles (EVs) currently available in the market have almost twice the carbon footprint during the manufacturing phase compared to conventional ICE vehicles even though their subsequent emissions during usage are practically negligible. And for the entire world to embrace EVs would still take a significant amount of time, especially the developing and the under-developed countries. This, in turn, means the pressure is back on the manufacturers of conventional ICE (and hybrid) vehicles to amp up their technologies to produce less damaging cars to the environment.

I guess what I’m trying to say is this: In a corporate world, where the general impression of people working in the auto industry, put forth by pressure groups, is that of a person who’s “sold their soul”, it is heartening to know that engineers like us, directly or indirectly, are making whatever minuscule efforts we can, in our capacity, to bring about a positive impact as far as the environment and the struggle of achieving climate-change goals is concerned.

Read more

Use case 2: Automatic species identification in the Kosterhavet National Park.

Figure 1: Kosterhavet National Park Map

The Kosterhavet National Park is one of Sweden’s most important and unique marine environments and is not only the first but also the most species-rich marine park in all of Sweden. The area obtained official protected status in 2009 and monitoring changes in the marine environment has become a top national priority. Understanding this complex ecosystem and its development in light of a warming planet and increased human activity is crucial to ensuring its survival for generations to come.

Initial problem formulation:

Researchers studying the changes in the marine ecosystem over the last 30 years at Kosterhavets National Park face many challenges, including:

· Storing and accessing observation data in a centralised and standardised way
· Identifying species when most of the captured footage contains little to no fauna/flora
· Analysing past footage with poor visibility conditions and/or low camera resolution

In the past, this footage has been examined and annotated manually by experts in marine biology. Given the advances in data science, this is simply not an efficient use of their time, and we believe that machine learning techniques could help to automate the parts of the process that significantly slow things down.

Initial Research Question

Could we set up the data infrastructure for a highly-performant object detection model to help us detect an important habitat-building marine species (Lophelia Pertusa) in real-time footage?

Figure 2: An example of Lophelia Pertusa

Data collection

Figure 3: ROV used for monitoring seafloor

For the last 30 years, the scientists at the Tjärnö Marine Laboratory have used Remotely Operated Vehicles (ROVs) and underwater cameras to monitor the area around the Kosterhavet National Park. The movies and images available to us today not only present an otherwise unseen part of the national park, but they also can take us 30 years back in time. In our research, we want to study how climate change and human activities influence the fauna in this area and which positive effects the protective status as a national park had on the seafloor habitats.

Data Annotation Workflow

Citizen science platforms allow researchers to educate others on topical issues such as marine biodiversity and environmental management, whilst also benefiting from crowd-sourced data annotation. In this project, we used the Zooniverse platform, which allowed us to upload the captured footage directly and present these as part of online workflows for citizen scientists.

The data annotation process consists of two separate workflows on Zooniverse. The first provides a 10-second video clip and then guides annotators to select the species they see in the clip with an extensive tutorial (as shown in Figure 4). Annotators may also provide the time of the first species appearance during the clip.

Figure 4: Screenshot from workflow 1 of Koster Seafloor Observatory on Zooniverse

Using the information from the first workflow, we can reduce the amount of footage significantly (by excluding all footage with no identified species) and extract relevant frames by species based on an aggregation of the citizen scientists’ input.

Figure 5: Screenshot from workflow 2 of Koster Seafloor Observatory on Zooniverse

The second workflow then prompts annotators to draw bounding boxes around the species they identified earlier (see Figure 5). These annotations are then aggregated and used to train the object detection model.

Data Infrastructure

Setting up the infrastructure for storage and retrieval of footage and metadata is a crucial part of ensuring the project’s longevity and scalability. For this purpose, we used a high-performance Linux-based project-specific server hosted by Chalmers University of Technology, Gothenburg. We also created an SQLite database to keep track of the information related to the movies and the classifications provided by both citizen scientists and machine learning algorithms. The database followed the Darwin Core (DwC) standards to maximise the sharing, use and reuse of open-access biodiversity data.

Automatic object detection model

The machine learning model chosen for this task is generally referred to as a single-shot detector, which means that the model looks at the entire image and predicts multiple objects inside the image at once, as opposed to region-based models which first identify regions of interest and then detect features in these regions as a second step. This leads to faster detections in most cases and allows these models to predict in near real-time. For our object detection model, we used the YOLOv3 architecture by Redmon and Farhadi (2018) [5]. As the name suggests, the third iteration of the YOLO model offers improvements. Still, it essentially uses the same building blocks as its predecessors, such as Darknet-53 as the feature detector, adding useful enhancements such as a multi-scale detector at three different scales to detect smaller objects. Our model was based on an open-source implementation to ensure that our results could be easily replicated and also be kept up to date with advances in future YOLO releases.

Figure 6: YOLOv3 model architecture

See the model in action (video)

Visualising our overall workflow

Figure 7: Flowchart showing the entire Koster Seafloor Observatory as an end-to-end process

Tools in development and education

In line with our goals at ODF to make our project outputs accessible and maximally flexible, we have created a web app that allows non-technical audiences to upload their own footage of Lophelia Pertusa corals and obtain predictions directly from the model. To increase understanding of the model, users can also tweak important parameters that affect model output, such as the confidence threshold and overlap thresholds of various bounding boxes.

Figure 8: Screenshot from the web app (to be released soon)

Code repositories

We open-source all our code in line with ODF objectives. Here are the repositories associated with the Koster seafloor observatory:

· Github repository for the machine learning model
· Github repository for the system data flow

Key Takeaways

· Our object detection model was able to analyse 150 hours’ worth of footage in just 30 hours.
· Our open-source database is now a valuable resource contains well-documented observation data for future research on marine environment management in this region and beyond.
· There is room to expand to other important habitat-building species as the model has been successfully validated on Lophelia Pertusa, and to continue adding footage as more protected areas get added.

The challenges ahead

No machine learning model is ever perfect, and as such, we still face many challenges that will be addressed in future iterations, including:
· Dealing with low-confidence or low-consensus predictions
· Improving underwater image quality for older footage
· Expanding the model to include more key species
· Improving model confidences by adding feedback from the model into future training data.








For more information on ODF and the work we do, visit the ODF Sweden website here.

Read more


The ROV we are using has a variety of sensors, such as IMU, magnetometer, barometer, ultrasonic sensors, and a camera. An HKPilot handles the sensor data while all the computations are, since last year, made on a Raspberry Pi also mounted on-board. This change increases the ROV’s autonomy, and the ethernet connection to the workstation is now only used to transmit commands to the ROV and to be able to watch the live feed from the on-board camera.

Several different autonomous modes have been implemented. It is, for instance, possible to follow objects, once the camera has found the object in question (such as a yellow ball). It will then be able to maintain the same distance to it as it moves. Furthermore, it is possible to position the ROV in the pool and to maintain position while being exposed to streams or other movements, e.g. by people. Of course, an operator can control the ROV by hand via the ethernet connection. With the wide range of sensors mounted in the ROV today, the door is open for future development of even more advanced autonomous modes.

The codebase, as of today, consists of a mixture of manually coded functions as well as generated code from Matlab/Simulink models. After the ongoing project, all code will hopefully be generated, which will give the development environment a lower threshold for new users.

Now, a deep dive (pun intended) into how the controller was updated during the last year. Previously, the control system of the ROV consisted of six different PID controllers, one for each degree of freedom (DOF). It resulted in poor performance when controlling multiple DOFs simultaneously. The reference generated only excited one DOF at a time, which made the pathfinding inefficient. For this reason, a Linear Quadratic Regulator (LQR) was implemented, and the system model was linearized.

The implemented LQR is a multiple-input-multiple-output (MIMO) controller which minimizes the criteria

which should be interpreted as minimizing the difference from requested states as well as the amplitude of the input signal. The model setup for the LQR is

The optimal linear state feedback controller, assuming known states x, is given by

where L is given by

The controller was implemented in Matlab/Simulink, as can be seen below. The central part is the matrix multiplication of the gain matrix L and the control error. The other crucial element is the feed-forward block that derives the input signals for maintaining the current attitude and depth of the ROV. The difference in result between the LQR and the PID control system can be seen in the picture below, although it is hard to visualize a 3D-path on paper.

Interested to know more about the project, contact Rikard Hagman at our Linköping office. You can also find more information and videos here:

Read more


Model predictive control (MPC) is the big brother in control theory. One common interpretation is that as long as one has access to good computation power MPC can control very complicated systems optimally. There is a bunch of reports on the world wide web that shows simple simulations of cars driving on roads, using MPC, which might lead one to think that this is a great way to operate selfdriving cars.  

We set out to find out just how accurate this claim is and the limitations MPC has within this field. At the heart of the MPC algorithm is the optimization problem. It is the job of the optimizer to calculate the best series of control inputs for the car in its given environment. This thesis has, therefore, evaluated three different algorithms. Two of them are Sequential quadratic programming (SQP) and Interior-point (IP) method, which are powerful and very common optimizers for nonlinear problems. The third optimizer is the Proximal Averaged Newton-type method for Optimal Control (PANOC), which is a new state of the art optimizer. 


The optimizers were tested in simulation where a road was modeled together with some obstacles. The Collision cone method was used as a collisionavoidance system, and it works by calculating the subset of relative velocities that the car can have, which will lead to collision. Then the optimizer choose the best velocity that is not within this calculated subspace. The collision cone only works if the obstacle is moving linearlyThus we developed our nonlinear version. This method calculates the predicted collision point, which is shown as a black dot in the figures below and applies the collision cone constraint on an obstacle projected onto it. 

All the optimizers managed to drive the track but with different results. In short SQP and IP are not performing very well in terms of computation speedbut SQP converges optimally. PANOC, on the other hand, converges very fast, around 33 ms per optimizer call, and is not as sensitive to disturbance. Unlike the others, PANOC can also be initiated from any arbitrary starting point and still manages to converge well. 

What follows is a recommendation of application for the optimizers with this thesis as a foundation for the rating. 

IP is recommended for lightly constrained problems. The environment should be a linear and smooth environment. If the solution is located close to or on difficult constraint borders, IP will have a hard time converging to a solution. 

SQP is best used in medium to highly constrained environments that are not timecriticalIf the algorithm can be given enough computational time, it will improve the quality of the solution and is, therefore, suitable for systems that are unstable or if the need for an optimal solution is high. For mildly constrained environments, the need for warm start is not that high. The algorithm can be used in nonlinear, non-smooth environments, given that it is warm started with a guess close to the solution. 

PANOC can be used in timecritical systems due to its superior convergence rate. It will rapidly provide a solution. It can also be used in highly constrained environments and even non-smooth and nonlinear ones. It should not be used in said complicated environments if the system is unstable or if the requirements on optimal output are high since it is not reliable. The output should be checked so that it meets the specified requirements. PANOC is a suitable choice when there is no way to warm start the optimizer with a good initial guess since it can converge fast from any initial state. 


Why PANOC is superior in terms of computation speed is due to that it is constructed with very cheap operations, and it cleverly combines the proximal operator with a quasi-newton method. These two methods together deliver an algorithm that is both fast and, thanks to the proximal operator, can converge in non-smooth highly nonlinear environments. PANOC exhibits a couple of strange behaviors, and the cause of these was not found. It does not converge optimally from time to time, and this behavior does not improve with increased computation time. PANOC is definitely a worthy candidate and will most probably find its way out in the market due to its superior advantage in the field of nonlinear non-smooth applications such as embedded MPC controllers for autonomous systems.  

In general, the environment that has to be modeled for a car to be able to drive in a realistic setting is to complex for current optimizers. MPC can not be used as a controller on its own. For it to give reliable results, the problem has to be simplified a lot so that more accurate response time and convergence can be guaranteed. One possible solution could be to split the optimization problem into two parts. For example, by having one optimization algorithm focusing on driving in its lane and avoid collisions, while another one determines when it’s possible to perform aovertaking maneuver. This could simplify the environment enough to allow the algorithms to behave more reliably.

Read more

An old and a new edition of Sympathy for Data

Combine is proud of the open-source history of Sympathy for Data and to be able to offer the product free of charge for everyone. A tradition we plan to continue in the form of an open-source edition of Sympathy for Data. With the introduction of Sympathy for Data 2.0.0, we decided to introduce a new edition, called Sympathy for Data Enterprise, a commercial product built on top of the open-source edition which extends the functionality with enterprise-tailored features and specialized toolkits. From the start, it includes a machine learning, time series and image analysis toolkit. The number of toolkits and the overall enterprise functionality will be extended with future versions.

We will offer the enterprise edition as a subscription-based license, including support, maintenance and continuous updates with new features and functionality. As usual, we are working in a customer-centric way to bring the best and most requested features for the new toolkits as well as the platform to our enterprise customers. To not forget our valued users of the open-source edition, we will also include selected features and improve functionality in this edition.

With the release of Sympathy of Data 2.0.0, we also changed to semantic versioning and a new release cycle which we will describe later. In general, releases of the enterprise edition will follow the releases of the open-source edition.

What is new in Sympathy for Data 2.0.0 and 2.1.0

Time Series Analysis Toolkit (Enterprise only)

The Time Series Analysis Toolkit provides a set of specialized nodes to work with time-series data and to perform signal processing, including dynamic time warping, time-series aggregation (PAA, SAX, APCA), time series transformation between different time and frequency domains, and more generic nodes for finding local min/max and plateaus.

Machine Learning Toolkit (Enterprise only)

The Machine Learning Toolkit provides nodes to describe, analyze, and model data. It provides nodes for features selection, dimensionality reduction and data pre-processing as well as supervised and unsupervised machine learning algorithms, including K-nearest neighbours, feed-forward neural networks, support vector machines, decision trees and ridge regression. On the advanced clustering methods side, the toolkit contains nodes for density-based spatial clustering (DBSCAN) and agglomerative clustering amongst others.

Image Analysis Toolkit (Enterprise only)

The Image Analysis Toolkit provides a set of specialized nodes for working with image data, including functionality such as property extraction from labelled images, stereo image analysis, generating noise as well as de-noising images, object identification using pre-trained Haar-cascades or simple template-matching.
Furthermore, the toolkit contains utility nodes for arranging images, converting figures to images, or converting between different data types in images and nodes for post-processing data, for example for selecting a relevant subset of lines from a Hough transform or matching key points identified in two images.

Indexed ADAF files

From version 2.1.0, we store a structure index in the ADAF files. It makes repeated queries of the ADAFs structure information much more efficient. For operations that primarily query the structure, for example, listing the available signals and their types, this leads to significant speedups.

Cursors for ADAF and Table viewer

In the latest version, we added cursors to the ADAF and Table viewer. It is a quick way to measure the difference, min, max, and mean values of the selected interval. The data between two cursors can also be exported as a CSV file for further analysis elsewhere.

Signal search for ADAF viewer

Version 2.1.0 adds a new view for searching for signals within an ADAF. It makes it more convenient and faster to locate signals when working with ADAF files containing many different rasters.

New figure wizards

With version 1.6.2, we introduced a new way of configuring new figure nodes with the help of a wizard. In version 2.0.0, we added new wizards for creating histograms, heatmaps, image, and timeline plots. Furthermore, the wizard view is now shown by default when configuring an empty figure node.

Undo in calculator

The powerful Calculator node has finally received an Undo/Redo functionality. It is accessible through buttons in the column toolbar as well as keyboard shortcuts. To make room for the extra buttons, the layout received some optimizations and shows each calculation in the list on a single row.


Sympathy for Data can now be installed without requiring elevated permissions (administrator) in all cases.


With the release of version 2.0.0, we changed to semantic versioning (major.minor.patch) with guaranteed backwards compatibility for minor and patch versions. The general plan is to release one major version and one to two minor versions per year.

If you want to know more about Sympathy for Data, its use cases, functionality and the different editions or you have comments and suggestions for the product, don’t hesitate to contact us at Combine or check out our webpage.

Read more

    Contact us