A novel approach combining importance sampling with DDPG for improved deep reinforcement learning performance and sample efficiency.
Advanced planetary simulation system leveraging Unreal Engine 5 and Chandrayaan-2 Terrain Mapping Camera data for high-fidelity lunar surface modeling and rover mission planning applications.
CogniPlan enhancement with cross-attention mechanisms and ViNT integration for semantic-aware adaptive exploration.
Deep reinforcement learning-based reactive planner for large-scale Lidar-based autonomous robot exploration in 2D action space with novel alpha conditioning for exploration-exploitation control.
3rd place autonomous drone navigation system using vision-based control with masking, ray-tracing, and closed-loop yaw control.
Monocular vision-based UAV treeline following system using instance segmentation and curve-fitting algorithms for precise orchard navigation under wind disturbances.
CogniPlan leverages conditional generative inpainting models to predict multiple plausible layouts, mirroring human cognitive maps for uncertainty-guided path planning in unknown environments.
Path planning in unknown environments is a crucial yet inherently challenging capability for mobile robots, which primarily encompasses two coupled tasks: autonomous exploration and point-goal navigation. In both cases, the robot must perceive the …
FEM-validated hockey ball launcher achieving 150 km/h with CFRP composite design and safety factor 2.07 at 496 RPM.
5th place + Best Maintenance Award - ExoMy rover navigation with UR3 arm manipulation using ArUco detection and MoveIt motion planning.