Welcome to the Estimation, Search, and Planning (ESP) research group, a member of the Oxford Robotics Institute (ORI). We motivate a knowledge-based approach to robotics research by taking on challenging, real-world problems. We do this to test our solutions and force us to understand the fundamental challenges of next-generation robots.
Please have a look around our website to find out more about who we are and what we do, including our publicly available code and datasets. You may also be interested in our Twitter, YouTube, and GitHub accounts.
Kevin has submitted a journal paper on Multimotion Visual Odometry (MVO) to IJRR. This is his definitive version of the work we're doing in ESP on measuring moving scenes from a dynamic camera. You can learn more from the trailer video and multimedia extensions on YouTube and the paper preprint on arXiv.
Marlin has submitted a journal paper on asymmetric bidirectional planning to IJRR. We're happy to provide its algorithms, Adaptively Informed Trees (AIT*) and Effort Informed Trees (EIT*), as open-source OMPL code. You can find out more in the trailer video and multimedia extensions on YouTube and the paper preprint on arXiv.
Marlin has been working on planning problems that seek to optimize the distance between paths and obstacles. This is a difficult problem for informed search algorithms, like A*, because there aren't very many effective and cost-efficient admissible heuristics for obstacle clearance. He's written up a short technical note on an admissible heuristic for clearance in search space that you can find on arXiv.