Small Business Innovation Research/Small Business Tech Transfer
Approximate Cartesian Control for Robotic Tool Usage with Graceful Degradation
Project Description
Many of NASA's exploration scenarios include important roles for autonomous or partially autonomous robots. It is desirable for them to utilize human tools when possible, rather than needing to build custom tools for each robot. Control of robotic manipulators for tool usage generally requires a very precise Cartesian-space trajectory of the tool tip (e.g., moving a marker along the surface of a whiteboard or rotating a screwdriver about an axis). Well-known techniques exist for manipulator control in Cartesian space, most of which necessitate solving a series of Inverse Kinematics (IK) problems. Closed-form IK solvers work well for 7-degree-of-freedom (DOF) arms with rigid tool attachments, but cannot handle non-rigid tools that slip in the robot's hands. Numerical IK approaches are more generic and can handle non-rigid links to tools, but can be slow to converge. More importantly, if any joints fail or become limited in their range of motion, the robot arm essentially becomes 6-DOF or lower. IK solvers often fail in these lower DOF spaces because the configuration space becomes non-continuous and full of "holes". As a result, a 7-DOF robotic arm in space might be rendered largely useless if a single joint fails or even loses mobility until it can be serviced. TRACLabs proposes to investigate an alternative approach to traditional Cartesian control approaches, which rely on complex IK solvers that go from Cartesian space backwards to joint space. We propose to leverage cheap memory and modern processing speeds to instead perform simple computations that go from joint space forwards to Cartesian space. Such techniques should overcome common changes to a manipulation chain caused by tool slippage or the grasping of a new tool and to overcome uncommon changes to a chain caused by joint failures, reduced joint mobility, changes in joint geometry or range of motion, or added joints.
More »
Anticipated Benefits
NASA's future robotics missions are expected to rely heavily on dexterous robots. These robots will need sophisticated control software in order to function. These robots will assist humans with tasks as well as work with no human presence to perform tasks such as assembly and terrain preparation. This work is directly applicable to current NASA robots such as Dextre and Robonaut-2, both of which are currently on-board ISS. Even under teleoperation, this software could overcome any failing or ailing joints to still get experiments performed or tasks completed. This work is also applicable to NASA ground robots such as Centaur and future exploration robots.
Dual arm manipulation robots are also becoming more common in industrial settings. Rethink Robotics and GM are making pushes for more generic humanoid robots to be use in factory settings. We believe the proposed technology will allow these robot to continue at least partial work if joint breaks during operation. The Department of Defense (DOD) is investing heavily in remote robotic operations including unmanned ground vehicles and is beginning to equip these vehicles with sophisticated manipulation systems. This manipulation systems are used for Explosive Ordnance Disposal (EOD), medical operations, entering and clearing buildings, moving supplies and unloading pallets. Our technology will greatly increase the usefulness of highly dexterous robots in military environments We expect substantial interest in the DOD to these kinds of technologies. We are also working with the US Army on remote medical robotics applications. In addition, we see applications in the urban search and rescue (USAR) arena and are coordinating with Texas A&M's Center for Robot-Assisted Search and Rescue. We are also investigating remote operation of robots on oil drilling platforms to reduce manpower and allow for continued operation in the face of storms that require evacuation of platform personnel. We are also investigating the automation of remotely operated underwater vehicles, such as those produced by Oceaneering. More »
Dual arm manipulation robots are also becoming more common in industrial settings. Rethink Robotics and GM are making pushes for more generic humanoid robots to be use in factory settings. We believe the proposed technology will allow these robot to continue at least partial work if joint breaks during operation. The Department of Defense (DOD) is investing heavily in remote robotic operations including unmanned ground vehicles and is beginning to equip these vehicles with sophisticated manipulation systems. This manipulation systems are used for Explosive Ordnance Disposal (EOD), medical operations, entering and clearing buildings, moving supplies and unloading pallets. Our technology will greatly increase the usefulness of highly dexterous robots in military environments We expect substantial interest in the DOD to these kinds of technologies. We are also working with the US Army on remote medical robotics applications. In addition, we see applications in the urban search and rescue (USAR) arena and are coordinating with Texas A&M's Center for Robot-Assisted Search and Rescue. We are also investigating remote operation of robots on oil drilling platforms to reduce manpower and allow for continued operation in the face of storms that require evacuation of platform personnel. We are also investigating the automation of remotely operated underwater vehicles, such as those produced by Oceaneering. More »
Project Library
Primary U.S. Work Locations and Key Partners
| Organizations Performing Work | Role | Type | Location |
|---|---|---|---|
| TRACLabs, Inc. | Lead Organization | Industry | Webster, Texas |
Johnson Space Center
(JSC)
|
Supporting Organization | NASA Center | Houston, Texas |
Primary U.S. Work Locations
-
Texas
Suggest an Edit
Recommend changes and additions to this project record.