- About the Center
- Team RIM
- Innovation and Industry
Recent Research Topics
Below is a sampling of recent research topics showcased at the previous yearly RIM poster sessions. A pdf each poster of the selected topics can be found below as well as the the full booklet.
The development of control applications for multi-agent robot and sensor networks is complicated by the heterogeneous nature of the system involved, as well as their physical capabilities or limitations. We propose a software framework that unifies these networked systems, thus facilitating the development of multi-agent control across multiple platforms and application domains.Pancakes A Software Framework for Robot and Sensor Applications brochure.pdf
Multi-robot SLAM systems are necessary to coordinate teams of robots by producing consistent, reliable maps of the environment. One challenge in a multirobot system not present in single robot SLAM is finding globally consistent labels for landmarks observed by separate robots when the starting reference frames of the robots are not known. We present a novel, RANSAC-based, approach for performing the between-robot data associations and initialization of relative frames of reference, obtaining an end-to-end multi-robot SLAM system, when combined with our previous DDF-SAM approach, for which have only shown simulated result until now.Fully Distributed Multi-Robot Simultaneous Localization and Mapping brochure.pdf
We present a feature based mapping technique that allows for the use of planar surfaces such as walls, tables, counters, or other surfaces as landmarks. These planar surfaces are detected in 3D point clouds, and provide measurements via their surface normal and perpendicular distance. We also map the convex hulls of the observed planar patches and use these for data association, allowing multiple non-overlapping coplanar landmarks to exist in the map. Maps of such planar surfaces could be useful for semantic mapping, and could benefit mobile manipulation tasks.Planar SurfaceMapping brochure.pdf
This project involves a rapidly iterating, user-integrated design process to produce a user interface and underlying autonomous capabilities, allowing a single quadriplegic user/collaborator to
physically interact with his environment through a PR2 robot. In addition, the user provides a case-study in the needs of the severely disabled with respect to potential robotic solutions. His insight into patient needs and capabilities alike allows for the design of an interface which is both effective and easy to use, and remarkably powerful as a results.
Future Naval Combat Operations and Systems will entail small expeditionary forces which must monitor and protect large and complex areas continuously. The purpose of HUNT is to push the state-of-the-art in complex, time-critical mission planning and execution for large numbers of heterogeneous vehicles collaborating with war fighters. Sophisticated cooperation among intelligent biological organisms will offer critical insight and solution templates for many hard engineering problems.HUNT Project Canid Hunting Behavior brochure.pdf
We present a robotic surveyor system that incorporates on-line measurements to dictate the path of navigation. GOAL: Ensure that the area of interest is properly surveyed and minimizes error
between estimated values and ground truth data.
In this paper, the whole-body motion control of a dynamically-stable two-wheeled humanoid robot is investigated. With feedback linearization as a control law cooperating with a stochastic optimization
technique called particle swarm optimization (PSO) to search for an optimal set of certain unknown parameters, an optimal robot motion can be realized according to a pre-fined performance index. The main contribution of this work is our first success on implementing the proposed control strategy on the actual physical system, Golem Krang, a novel two-wheeled humanoid robot developed at Georgia Tech. The experiments are demonstrated by two primitive motions, standing from the ground and deceleration.
Robots interacting with humans can benefit from communicating in a manner that is socially relevant and familiar to their human partners. Believable behavior establishes appropriate social expectations. Our work addresses the overall problem of how to generate believable or human-like motion for an anthropomorphic robot. Anthropomorphic robot bodies are different from human bodies.
The challenge is to create human-like motion for robots autonomously from minimal input information, despite the differences between robots and humans. Our work focuses on a series of algorithms that work toward improving motion quality for social robots.
A full listing can be found here in the yearly poster session booklet.