Autonomous systems are always-on, always connected, and their designs are never complete. They require Continuous Integration and Continuous Development (CI/CD) just to operate effectively. Today, a typical connected vehicle will generate over 25GB of data per hour. Can your desktop software seamlessly ingest and process 25GB+ of data per hour to retrain your neural networks? Collimator can.
Collimator is a modeling and simulation platform for engineers to design and test complex, mission critical systems in a way that is reliable, secure, fast and intuitive. Collimator provides a unified environment to design, simulate, test, and continuously upgrade autonomous systems in a world where big data is continually used to train neural networks and engineering design continues through the lifecycle of a system. Reduce your risk and bring products to market faster by using Collimator to:
Design autonomous systems using a graphical UI or Python Notebook
Integrate with 3D scenario modeling tools to get a system level view of performance
Use AI and ML tools to generate synthetic data or model autonomous systems
Run millions of test cases using HPC in the cloud before implementing in hardware
Call external libraries and models like Pytorch and Tensorflow directly within Collimator
Compare real-world operation with virtual operations in real-time, continuously
Model your autonomous systems the way that you are most comfortable using our no code block diagram environment or our Python environment
Import trained neural networks from open source AI and ML packages (e.g., Tensorflow and Pytorch) to seamlessly generate synthetic data or simulate autonomous systems
Call functions from any open source library (e.g., Pytorch, Tensorflow, Keras) or run your own external code directly within Collimator
Seamlessly send data back and forth between 3D modeling tools such as CARLA and AirSim to get a system level view of performance
Automatically run HPC simulations in parallel on the cloud with one click and without worry that system memory will be a bottleneck
Import data into Collimator, compare the real-world response to the simulated data, export data back to your data warehouse for storage or to your neural network for additional training
Collimator is Python-based, so you have access to open-source libraries like Numpy, Tensorflow, and Pytorch. Convert Python code into C automatically using our system designer software
Our software design tool’s High-Performance Compute (HPC) lets you run millions of test cases, test various scenarios, and simulate conditions difficult or expensive to reproduce in the real world
Design system software in multi-player mode, with no more miscommunications. Collimator increases output by 10x when you’re collaboratively designing various elements of a large system
We have simple, easily understood plans for users ranging from small teams to large enterprises. With our free plan, students can try Collimator at no cost
Because our software designing tool is built in Python, your users can easily import AI and ML tools like Pytorch, Thane and Keras and use them to test and perfect your system design
The more data you have, the better. Our system designer software uses data from various sources to represent situations that will inform and improve your overall system
Software systems design is the process of outlining and building the different components of a software while ensuring that data is sent and received efficiently through the software as a whole.
System design tools provide an interface for system designers to manage their systems in a standardized way. Ideal system design tools will feature intuitive component drag & drop, easy team collaboration, management of version control and seamless integration with developer tools.
There are many apps that users turn to to meet their system design needs. System Composers by Matlab/Simulink, Ansys SCADE, and Collimator are all great tools for system design across various fields.
.png){kind=small}
.png){kind=small}
.png){kind=small}
.png){kind=small}
.png){kind=small}
.png)
