The Barrett WAM™ arm
Imagine the following on-orbit scene: the thrusters of a free-flying robotic platform, carrying a robotic arm tasked with repairing the Hubble telescope, are pushing it gently toward its target task.
Now consider two mission-design alternatives: the robotic arm is (1) conventional or (2) a WAM™.
Alternative 1: A conventional robotic arm is outfitted either with active proximity-sensing skins designed to prevent collision, with force/torque sensors designed to minimize contact force, or both. Proximity sensors create an immediate paradox: the whole reason to have a robotic arm is to allow flexible manipulation of the task. But the proximity sensor that protects against physical contact inherently limits flexibility by prohibiting contact tasks.
Force/Torque sensors do allow tasks to occur, but rely on a two-step process that is prone to accident. When there is physical contact, sensors rapidly report the sudden presence of forces/torque, however, the inertia of even the lightest motors and robot structures take many 10s of milliseconds to respond to commands to begin to reverse direction. Even if contact occurs at only 0.1 m/sec, the most responsive robotic arm will have moved several millimeters, with the risk of serious damage that this implies.
With a conventional robotic arm dependent on Force/Torque sensors, the damage is done before there is time to react. This inherent delay in response also implies serious damage if a positioning error causes the conventional robot arm to be driven into the Hubble.
Alternative 2: Use the naturally-backdrivable WAM arm. The WAM is highly responsive to contacts (even when its power is cut), not just at its endtip but all across its link surfaces. Also the WAM does not rely on any active force or torque sensor, but rather controls motor current precisely. Rather than the two step closed-loop process of (1) sense force and (2) move away from force, the WAM arm commands force directly. Purposeful, controlled task forces are robustly controlled. If power is lost, the WAM’s fail-safe condition is lightly-damped backdrivability. If a platform-positioning error causes a collision, then the WAM softens any contact between Hubble and the thruster platform. Impact energy is passively dissipated through the backdriven motor-winding resistances.
Lightest, Safest, and Ultra Power Efficient
The Barrett WAM™ arm
The WAM has hidden technology (patents pending) codenamed “Puck” servoelectronics that validate its title as the lightest and most power-efficient full-size robotic arm in the world. The Puck will enable electronics for future extra-terrestrial rover missions to be more robust, compact, and efficient while enabling expanded arm functionality and other rover servodrives.
A conventional robot is not just what you see in a product photo. It also includes a heavy metal box called the controller cabinet which contains power supplies, servoamplifiers (one per brushless motor), and a high-level trajectory controller. By convention, the mass of the controller box is conveniently excluded from the robot’s rated mass, even through the two are inseparable. Some new prototype systems attempt to squeeze some of the controller-box electronics into the robot structure, but with a proportional increase in wiring complexity, thermal cycling, and component failure.
In contrast, the WAM™ system has eliminated these conventional electronics, and replaced them with solid-state “Pucks,” 43-gram bottle-cap sized hermetic modules (one attached to the back of each brushless servomotor and sharing a 4-wire bus). Each Puck includes a precision position encoder; precision current (i.e. torque) sensing; a 32-bit DSP supporting distributed trajectory control; battery-native power conditioning, space-vector electronic commutation; efficient current amplification; and auxiliary user features such as strain-gage amplification and general-purpose inputs/outputs (GPIO).
The Puck modules are so power efficient that a 4-dof WAM operating against full gravity draws only 28 Watts. WAM™ on-orbit application, in the absence of gravity, will reduce continuous consumption to only 12 Watts. Low power consumption translates directly to less mass that must be placed on orbit: reduced battery storage mass (required for operation during orbital nights) and fewer or smaller solar arrays.