Parker Ewen

We incorporate both visual and tactile information to improve semantic maps for robots.

We develop a modular neural network for real-time semantic mapping in uncertain environments.

This paper proposes a Bayesian inference framework for real-time semantic mapping that uses RGB-D camera data to estimate terrain properties and their probability distributions, outperforming state-of-the-art methods and successfully predicting terrain properties in both simulated and real-world environments.

This paper introduces STORMMAP, a real-time planner for legged manipulators that optimizes center-of-mass motion and manipulation force trajectories to ensure dynamic stability and task completion, running significantly faster than existing methods by assuming pre-determined contact locations.