Origins

PI: Thomas Bewley

                 

The inspiration for the genesis of our lab arose from a final exam in an MS-level class in dynamics & control taught by the PI in 2003. The exam focused on a vehicle which could self transform between three primary modes of operation: horizontal roving, vertical roving (similar to that of a Segway), and pogo-stick-like hopping. The proposed vehicle has two large main wheels and a third, smaller, castoring wheel at the far end of a leg passing through the center of mass. It can steer in both roving modes by differential actuation of the main wheels, and can self upright by torquing the main wheels backwards. The first two exam questions focused on:

(a) the continuous-time finite-horizon optimization techniques necessary to plan an efficient righting maneuver, and
(b) the continuous-time infinite-horizon control techniques necessary to stabilize the upright roving mode.

By releasing the energy of a pretensioned spring within the leg while in upright roving mode, the vehicle initiates a controlled hopping motion. While airborne, the vehicle uses the two main wheels, in addition to two smaller wheels mounted orthogonally between the two main wheels, as "reaction wheels": when these wheels are torqued in one direction, there is an equal-and-opposite reaction torque on the vehicle. The final two exam questions focused on:

(c) the continuous-time nominally-time-periodic control techniques necessary to reject disturbances (to keep the vehicle oriented vertically to hop in place, or to reorient the vehicle to any desired configuration while in flight to hop sideways), and
(d) the discrete-time control technniques necessary to determine the desired configuration of the leg in preparation for each hop (thus facilitating a repetitive hopping motion that can be used to move from one point to another).

Perhaps the most interesting maneuver that such a vehicle can perform is the running "single hop". During this maneuver, the system recompresses its spring upon landing and returns immediately to upright roving mode, ready to conduct another hop when needed; if the vehicle has a large forward velocity in upright roving mode before the hop, it will move horizontally a significant distance while airborne. The resulting "pit-of-fire leap" is potentially useful - if deployed for exploration of a burning building, the vehicle might need to move quickly over a burning obstacle which could otherwise damage it. Further, hopping is inefficient, and should be executed only when the mission calls for it; for efficiency, the vehicle should roll whenever possible, either in upright or horizontal roving mode. In fact, maximally leveraging the efficiency of rolling motion is a common theme in many of the systems our lab considers.

Shortly after the exam mentioned above, three of the top UCSD controls students expressed interest in exploring how to build such vehicles, and the UCSD Coordinated Robotics Lab was born (see iHop v.1). To begin, our "robots" were nothing more than advanced dynamics & control experiments; it soon became evident, however, that the vehicles being developed were capable of much more versatile maneuvers than were traditional UGVs, and that these maneuvers could ultimately be useful in a variety of practical applications.