Abstract: The manipulation of unknown objects in
unknown environments remains a central challenge in the field of
robotics. Rapid and robust solutions are needed if we are to transform
robots from tools to teammates. Therefore, the goal of this work is to
develop principles for combining control algorithms and techniques with
hardware design for efficient, autonomous mobility and manipulation in
a rapid and robust way. Three projects in this area will be
highlighted. First, the ability of a mobile robot to autonomously
self-right on planar surfaces will be explored through the lens of a
configuration space map. Created off-line, this map can be converted
into a graph for rapid on-line path planning. Second, an approach
utilizing compliance for rapid autonomous door opening will be
described. Here, we turn the manipulation problem on its head,
treating the door hinge as the base of an open chain serial
manipulator, with the mobility platform as the end effector. Finally,
the development of a self-sealing suction cup array for rapid
grasping will be presented. The goal of this work is to simultaneously
reduce the amount of planning time required to obtain a robust grasp
while also expanding the range of object shapes and sizes able to be
grasped by a single end effector. In general, manipulation is
a capability with the capacity to greatly expand a robot’s utility, and
so this work attempts to cover the rapid manipulation of the self,
constrained objects, and free objects.