Announcing Collimator: Modeling and Simulation built for the Cloud

I started my career as an electrical engineer and made the transition to software development working at companies such as Motorola and Samsung. During that transition, I realized that software engineering tools were a lot more user-friendly, flexible, and built for scale than the tools I had been accustomed to. By contrast, the tools I used as an electrical engineer seemed to lack innovation and were made by companies founded in the 1970s and 1980s such as Cadence, National Instruments, and Mathworks. I thought that it would only be a matter of time until this changed. Yet today these tools continue to dominate the landscape for hardware development while on the other hand, it feels like almost every day there is a new and improved tool built by developers for developers!

Take MATLAB for example. It was the very first language I learned as a student. Together with its graphical extension Simulink, it is the de facto standard for Modeling and Simulation (M&S) in a wide range of industries. However, this desktop based suite still looks and feels the same as when I first used it 2 decades ago. It is also a closed environment, making compatibility with popular languages such as Python challenging.

My co-founder and I decided to start Collimator to bring the same level of innovation we had seen in software to electrical and mechanical engineering problems. Specifically, we aimed to improve the hardware development experience focusing on Modeling and Simulation. M&S is becoming more and more important for many companies because it helps reduce development risk and costs. Numerous studies have shown that the more and the earlier in their product lifecycle companies simulate, the more innovative their products will be!

The need for M&S is growing rapidly, especially in emerging technologies. Engineers are increasingly asked to develop electromechanical systems that are more complex, with shorter development times, and the tools must keep up from both a sophistication and performance level.

For instance, as engines become electrified, how do engineers rapidly optimize battery and motor performance? As devices become more connected with 5G and IoT, how do engineers efficiently deploy networks and communications? As machine learning and digital twins become more ubiquitous, how do engineers easily integrate these new technologies into their workflows? As multi-domain systems become more complex, how do engineers accurately understand the interactions between them?

We do not believe the solution is yet another programming language that engineers have to learn. We also do not believe that the ideal solution is a remote desktop approach to porting existing applications to AWS/GCP/Azure – which many incumbent companies loosely advertise as their “cloud” offering. These duct taped solutions turn out to be just as cumbersome and costly to set up, they require expertise to maintain, and the results do not provide the inherent benefits of a cloud-native app built from the ground up.

We believe the answer lies very close to what Figma did for designers. Like Figma, we are abstracting away the complexity from users and offering the simplicity and convenience of Python, so they can focus on what they do best: building amazing products.

With Collimator, users can open a web browser (no download or install required) and quickly draw block diagrams of their dynamical systems by dragging and dropping components. Users need only specify the parameters for each block and the connections between blocks. Users can write custom code using our tightly integrated Python environment, import existing modules, and run simulations programmatically.

Simulations are executed seamlessly in the cloud with a built-in High Performance Compute (HPC). Users can easily incorporate machine learning models trained with Pytorch or Tensorflow, as well as process data from real-world deployment. Users can vary key parameters and inputs of their systems, run multiple scenarios in parallel, and then collect data for post processing, analysis, and visualization. This allows the user to easily explore “what if questions” and test conditions that are difficult or expensive to reproduce in the real-world.

We are very excited about the prospects of simulation-driven product development. We believe that by empowering engineers with a next generation modeling and simulation platform, we can help accelerate the development of physical products that will make our lives better.

Reda Dehy