Probability theory is a framework for representing uncertainty – due either to variability or unmodelled, unobserved factors – in a multitude of observed quantities. Probabilistic modeling enables robustness to noise and uncertainty in a number of autonomous systems, and through the widespread adoption of probabilistic algorithms. Probabilistic modeling was responsible for robotics transitioning from fragile, hand-tuned systems, to systems that are robust to operating in the natural world with noisy sensors. In order to interact with a complex and not-perfectly-known environment, both robots and biological creatures need to represent and manipulate probabilities. While probability-based mathematical formalisms such as Bayesian Optimization exist, they often assume an unlimited amount of computing power, and it is unclear how they can be mapped into more realistic physical hardware, or into artificial neural networks. In recent years, a variety of methods have been developed to start to connect probabilistic information with neural networks. The overall goal of the project is to further develop these techniques and to apply them to the flexible soft-robotics systems
This record is a historical disbursement from Canada's open grants database — it documents a past payment, not an active program accepting applications.
View disbursement record →