Applying Feed-Forward Compliance to the Control of Electric Motors Used in the Joints of Walking Planetary Robotic Explorers
- Paper number
IAC-05-A3.P.08
- Author
Mr. Gregory Scott, University of Surrey, United Kingdom
- Coauthor
Dr. Alex Ellery, Surrey Space Centre, University of Surrey, United Kingdom
- Coauthor
Dr. Eddie Moxey, University of Surrey, United Kingdom
- Year
2005
- Abstract
Although the scientific return on the MERs has been spectacular, there is a fundamental problem with planetary robotic missions to-date: the locomotion system. The MERs are very advanced wheeled robots, capable of climbing reasonable-sized rocks and avoiding larger ones. However, they were designed to traverse regions only covered by smaller rocks and sparsely covered by larger rocks. Traversing regions with more, larger rocks would be difficult, if not impossible. Fortunately, a solution exists for the locomotion systems to negotiate these more hostile terrains: Legs. We are currently working on a European Space Agency contract to investigate the benefits of biomimetics to planetary robotic explorers. Specifically, we have been investigating the benefits of feed-forward compliance applied to each joint of a 3-DOF leg on a 6-legged walker through proportional derivative motor control and its impacts to the Drawbar Pull metric involved in Bekker Theory. Bekker analysis provides a basis for determining the theoretical performance of a vehicle traversing on a particular terrain to help define its trafficability. Rather than attempt to fully model an artificial muscle, it is the intention of this research to extract the principles of animal locomotion and adapt them to the control of electric motors for driving the legs. This is because artificial muscles are still in an immature state of development and certainly do not have the heritage expected in most space-rated vehicles. The target vehicle is a hexapod (6-legged) with 3 degrees-of-freedom (DOF) in each leg, modelled as a 2-DOF hip and a 1-DOF knee. Compliance is the ability of an object to yield elastically when a force is applied. The mechanics of a robotic leg and the equations required to move a leg from one place to another are well known. For example, the overall motor torque with the effects of compliance modelled can be determined through adding a feed-forward compliance factor. The investigation of compliance within an electric motor system has been far less studied. This project investigates three compliance models: First, the feed-forward compliance function to model a simple spring-mass-damper system, such as Hooke’s Law, is a simple replacement of displacement with the difference between joint equilibrium position and the deflection from equilibrium. Second, a sigmoid function is used to help a system converge to a desired output by limiting extreme deflections. Mathematically speaking, this is simply a special case logistic function that is asymptotically bounded on both extremes of displacement. Third, a biological example using Hill’s visco-elastic model will be used to emulate muscle compliance. Hill’s model is one of the best-known relationships concerning muscle action and is the hyperbolic curve that describes the dependence of force on velocity of movement. This research is expected to show that legged locomotion with a compliance model is more expedient than legged locomotion without compliance. It is hoped that the incorporation of muscle-like behaviours into electric motors will enable legged robots to better cope with the loose regolith and large obstacles found on Mars. This will be determined based on the measurement of the forward tractive effort of the vehicle (Drawbar Pull). An increase in forward motion across the soil will prove an improvement in Drawbar Pull, thereby showing the positive impact of compliance on the system.
- Abstract document