SW StarLab: Recognition, Action, and Interaction Algorithms for Open-World Robot Service
The goal of this project is to develop algorithmic software to enable service robots to operate effectively in unstructured open-world environments. Taking an open-world service perspective, our aim is to identify key academic, technical, and functional challenges and conduct research on the underlying technology to address them. Specifically, we focus on studying foundational technologies related to recognition, action, and interaction systems for robot services. The academic and SW results derived from this project will be released as open-source. We hope this will contribute to solving the challenges of open-world robotics. 🙂
Software
- Spherical Hashing [Project / Code(C++) / Code(MATLAB) / GitHub]
- Adaptive Rendering based on Weighted Local Regression [Project / Code]
- Memory-Efficient NBNN Image Classification [Project / Code / GitHub]
- Super Ray-based Updates for Occupancy Maps [Project / Code / GitHub]
- Kinodynamic Comfort Trajectory Planning for Car-like Robots [Project / Code / GitHub]
- Regional Attention Based Deep Feature for Image Retrieval [Project / Code / GitHub]
- Single Image Reflection Removal with Physically-Based Training Images [Project / Code / GitHub]
- Weakly-Supervised Contrastive Learning in Path Manifold for Monte Carlo Image Reconstruction [Project / Code / GitHub]
- Confidence-Based Robot Navigation Under Sensor Occlusion with Deep Reinforcement Learning [Project / Code]
- Part-based Pseudo Label Refinement for Unsupervised Person Re-identification [Project / Code / GitHub]
- Diffusion Probabilistic Models for Scene-Scale 3D Categorical Data [Project / Code / GitHub]