Lessons Learned from All-Terrain Hex-Limbed Extra-Terrestrial Explorer Robot Field Test Operations at Moses Lake Sand Dunes, Washington

Abstract

The Jet Propulsion Laboratory (JPL) is developing the All-Terrain Hex-Limbed Extra-Terrestrial Explorer (ATHLETE) as part of NASA's Exploration Systems Mission Directorate, Exploration Technology Development Program (ETDP). The program develops technologies for surface mobility and equipment handling, human-system interaction, and lunar surface system repair, and constructs dexterous robots and autonomous rovers that can drive over rough terrain and help crew explore, assemble, and maintain a lunar outpost. ETDP sponsors a series of field tests at lunar analog test sites where prototype robots can operate in ways that simulate lunar surface conditions. In this paper, we describe the lessons learned about ATHLETE operations at the most recent lunar analog field test in June 2008 at Moses Lake Sand Dunes, Washington. The Moses Lake field test was structured as a series of "acts" which correspond to unpiloted and piloted missions to the lunar surface in the 2019 to 2022 timeframe. The field test took place over a period of two weeks and involved several robots from various NASA field centers, including the Chariot lunar truck from Johnson Space Center, the K10 planetary rover from Ames Research Center, and ATHLETE from JPL. Lessons learned from the Moses Lake field test will be incorporated into the evolving design of the ATHLETE operations system, and will be tested at subsequent field trials.

Operating High-DoF Articulated Robots Using Virtual Links and Joints

“…With 36 revolute joints in six identical limbs, ATHLETE can be applied in a broad range of tasks. Often, an intended motion is easily visualized, but expressing it could be tedious-even using custom-built direct manipulation hardware [22]. Figure 1 shows one example of how virtual articulations can help: the operator would like to tilt the field of view of an un-actuated camera by rotating the robot's body about a virtual axis in space, with the legs moving as necessary.…”

Using Virtual Articulations to Operate High-DoF Inspection and Manipulation Motions

“…All show the ability of our mixed real/virtual interface system to help design specific motions which are rapid for human operators to conceptualize but difficult to express in prior operations interfaces, including several other software systems under development within NASA [4,11] as well as our own TRACK device used alone [6].…”

“…Next we show several inspection and manipulation tasks on the ATHLETE hardware that would have been challenging with prior operator interfaces, including an experiment where we combine our virtual articulation interface with a direct-manipulation input device that mimics one ATHLETE limb. We developed this device, the Tele-Robotic ATHLETE Controller for Kinematics (TRACK), in prior work [6]. We conclude by summarizing known limitations of our approach and possible future directions.…”

Motion learning in variable environments using probabilistic flow tubes

“…While manually commanding ATHLETE through complex tasks can be extremely tedious, our approach will allow operators to teach the robot desired motions. While teaching, an operator uses an interface device called the Telerobotic ATHLETE Controller for Kinematics (TRACK) [1] to teleoperate the robot.…”

“…While manageable for simple robots, this approach is tedious when controlling robots with many degrees of freedom, such as NASA JPL's ATHLETE (All-Terrain Hex-Limbed Extra-Terrestrial Explorer), which has 36 independent joints [1]. Currently, operators can either use low-level joint angle commands or choose among a handful of pre-programmed higher-level tasks.…”