Title: Multi Robot Object-based SLAM
Date: Tuesday, November 8th, 2016
Time: 11:00am to 1:00pm (EST)
Location: Klaus 1116E
The use of multiple cooperative robots has the potential to enable fast information gathering, and more efficient coverage and monitoring of large areas. In particular, multi robot SLAM, i.e., the cooperative construction of a model of the environment explored by the robots, is fundamental to geo-tag sensor data (e.g., for pollution monitoring, surveillance and search and rescue), and to gather situational awareness. For military applications, multi robot systems promise more efficient operation and improved robustness to adversarial attacks. In civil applications (e.g., pollution monitoring, surveillance, search and rescue), the use of several inexpensive, heterogeneous, agile platforms is an appealing alternative to monolithic single robot systems.
In this proposal, I aim at designing a technique that allows each robot to build its own object level map, while asking for minimal knowledge of the map of the teammates. In particular, I make the following three major contributions:
1. I present a distributed algorithm based on Distributed Gauss-Seidel to estimate the 3D trajectories of multiple cooperative robots from relative pose measurements. This approach has several advantages. It requires minimal information exchange, which is beneficial in presence of communication and privacy constraints. It has an anytime flavor: after few iterations the trajectory estimates are already accurate, and they asymptotically convergence to the centralized estimate. The DGS approach scales well to large teams, is resistant to noise and it has a straightforward implementation. I test the approach in simulations and field tests, demonstrating its advantages over related techniques.
2. I present an approach for Multi Robot SLAM which uses object landmarks in a multi robot mapping framework. I show that this approach further reduces the information exchange among robots (as compared to feature based DGS), results in compact, human understandable map, and has lower computational complexity as compared to low level feature based multi robot mapping.
3. Finally, I propose to extend the previous work to the case where object models are previously unknown and are modeled jointly with Multi Robot Object based SLAM.
Prof. Henrik Christensen (Advisor), School of Interactive Computing, Georgia Institute of Technology
Prof. Frank Dellaert (Co-Advisor), School of Interactive Computing, Georgia Institute of Technology
Prof. James M. Rehg, School of Interactive Computing, Georgia Institute of Technology
Prof. Patricio Vela, School of Electrical and Computer Engineering, Georgia Institute of Technology
Prof. John Leonard, Department of Mechanical Engineering, Massachusetts Institute of Technology