NeuroMotive exploits a number of unique functions provided by the NeuroRobotics Platform (NRP) including development and simulation of high-performance neural controllers, deep learning to support and automate the process for car modeling and neuro-controller tuning. The NRP supports physically simulated vehicles, manned or unmanmned, and environments, with a variety of NN engines (NEST, TensorFlow, etc.). The physical simulation supports embodiment of the neural models, empowering them to learn not in a vacuum but in relation to a rich, realistic environment, bringing Brain Science to Automotive Innovation.
The NRP developed by the Human Brain Project (HBP) is an integrative framework for in silico experimentation that allows the use of cutting-edge neural-based dynamical physics models in complex environments. The plasticity of these models and of the neural controllers is leveraged to facilitate and standardize machine learning processes for model learning automation and controller tuning. The simulation environment offered by the platform allows to meaningfully expand the scope of testing, including safety-critical situations that would prove undesirable or impossible to explore in typical testing environments.