DeepSim 2.0 - Deep learning based platform to generate superior controller software, faster
Updated: May 22, 2020
Minds.ai is proud to present the availability of the second version of the DeepSim platform. The second version improves upon the strengths of the first version, takes advantage of the latest cloud technology innovations and is suitable to a broader range of application verticals. The platform is already reducing time to market and improving performance for R&D and production development departments of select customers and is now rolling out to wider availability. See the below for more information.
The DeepSim platform originates from our years of experience in deep learning consulting and incorporates all the customer requirements and tools developed during those years. DeepSim is a versatile platform which can be used to automatically generate and optimize controller software for any problem where a simulator is available. In the absence of a simulator, the platform can also use historical data about the environment where the controller operates. Hardware in the Loop (HIL) systems that are robust enough to allow for some trial and error can also be effectively used to train the controller.
The need for this platform comes from the convergence we are witnessing of the following technology trends:
Autonomous driving: land, water and sky.
Increasing complexity: design and performance requirements and simulation capabilities.
Smart environments: intelligent edge devices enabling automation of homes to factories and cities.
Big data: availability of big data from multitude of sensors and large networked systems
These technology trends are driving the need for AI solutions which can perform sophisticated decision making, acting autonomously and often in real time.
Although reinforcement learning as a method has been around for decades it only now becomes possible to design controllers for these highly complex and interactive environments. The amount of compute required to train the networks and simulate the environments has only recently become feasible due to the emergence of cloud computing and hardware accelerators like GPUs. DeepSim has been designed from the ground up to take advantage of these emerging technologies and allows our customers to scale from small scale experiments to large scale production training using the DeepSim platform. With DeepSim you are no longer bound to the compute power of your laptop or workstation but can effortlessly run the training and analytics on your on-prem cluster or on your cloud subscription.
With this platform we enable our customers to develop neural network agents to generate control programs and optimize their assembly line design by taking advantage of modern advances in AI and abundant compute resources. The platform results in guaranteed superior performance vs. controllers designed by experts with a faster time to market.