Simon Chamorro

Kasra Rezaee, Peyman Yadmellat, Simon Chamorro

Motion planning under uncertainty is one of the main challenges in developing autonomous driving vehicles. In this work, we focus on the uncertainty in sensing and perception, resulted from a limited field of view, occlusions, and sensing range. This problem is often tackled by considering hypothetical hidden objects in occluded areas or beyond the sensing range to guarantee passive safety. However, this may result in conservative planning and expensive computation, particularly when numerous hypothetical objects need to be considered. We propose a reinforcement learning (RL) based solution to manage uncertainty by optimizing for the worst case outcome. This approach is in contrast to traditional RL, where the agents try to maximize the average expected reward. The proposed approach is built on top of the Distributional RL with its policy optimization maximizing the stochastic outcomes' lower bound. This modification can be applied to a range of RL algorithms. As a proof-of-concept, the approach is applied to two different RL algorithms, Soft Actor-Critic and DQN. The approach is evaluated against two challenging scenarios of pedestrians crossing with occlusion and curved roads with a limited field of view. The algorithm is trained and evaluated using the SUMO traffic simulator. The proposed approach yields much better motion planning behavior compared to conventional RL algorithms and behaves comparably to humans driving style.

Simon Chamorro, Jack Collier, François Grondin

We propose a novel low-complexity lidar gesture recognition system for mobile robot control robust to gesture variation. Our system uses a modular approach, consisting of a pose estimation module and a gesture classifier. Pose estimates are predicted from lidar scans using a Convolutional Neural Network trained using an existing stereo-based pose estimation system. Gesture classification is accomplished using a Long Short-Term Memory network and uses a sequence of estimated body poses as input to predict a gesture. Breaking down the pipeline into two modules reduces the dimensionality of the input, which could be lidar scans, stereo imagery, or any other modality from which body keypoints can be extracted, making our system lightweight and suitable for mobile robot control with limited computing power. The use of lidar contributes to the robustness of the system, allowing it to operate in most outdoor conditions, to be independent of lighting conditions, and for input to be detected 360 degrees around the robot. The lidar-based pose estimator and gesture classifier use data augmentation and automated labeling techniques, requiring a minimal amount of data collection and avoiding the need for manual labeling. We report experimental results for each module of our system and demonstrate its effectiveness by testing it in a real-world robot teleoperation setting.

Martin Weiss, Simon Chamorro, Roger Girgis et al.

Millions of blind and visually-impaired (BVI) people navigate urban environments every day, using smartphones for high-level path-planning and white canes or guide dogs for local information. However, many BVI people still struggle to travel to new places. In our endeavor to create a navigation assistant for the BVI, we found that existing Reinforcement Learning (RL) environments were unsuitable for the task. This work introduces SEVN, a sidewalk simulation environment and a neural network-based approach to creating a navigation agent. SEVN contains panoramic images with labels for house numbers, doors, and street name signs, and formulations for several navigation tasks. We study the performance of an RL algorithm (PPO) in this setting. Our policy model fuses multi-modal observations in the form of variable resolution images, visible text, and simulated GPS data to navigate to a goal door. We hope that this dataset, simulator, and experimental results will provide a foundation for further research into the creation of agents that can assist members of the BVI community with outdoor navigation.