This paper explores the applicability of a Linear Quadratic Regulator (LQR) controller design to the problem of bipedal stance on the Minitaur  quadrupedal robot. Restricted to the sagittal plane, this behavior exposes a three degree of freedom (DOF) double inverted pendulum with extensible length that can be projected onto the familiar underactuated revolute-revolute “Acrobot” model by assuming a locked prismatic DOF, and a pinned toe. While previous work has documented the successful use of local LQR control to stabilize a physical Acrobot, simulations reveal that a design very similar to those discussed in the past literature cannot achieve an empirically viable controller for our physical plant. Experiments with a series of increasingly close physical facsimiles leading to the actual Minitaur platform itself corroborate and underscore the physical Minitaur platform corroborate and underscore the implications of the simulation study. We conclude that local LQR-based linearized controller designs are too fragile to stabilize the physical Minitaur platform around its vertically erect equilibrium and end with a brief assessment of a variety of more sophisticated nonlinear control approaches whose pursuit is now in progress.
This paper presents preliminary results of a semi-autonomous building exploration behavior using the hexapedal robot RHex. Stairwells are used in virtually all multi-floor buildings, and so in order for a mobile robot to effectively explore, map, clear, monitor, or patrol such buildings it must be able to ascend and descend stairwells. However most conventional mobile robots based on a wheeled platform are unable to traverse stairwells, motivating use of the more mobile legged machine. This semi-autonomous behavior uses a human driver to provide steering input to the robot, as would be the case in, e.g., a tele-operated building exploration mission. The gait selection and transitions between the walking and stair climbing gaits are entirely autonomous. This implementation uses an RGBD camera for stair acquisition, which offers several advantages over a previously documented detector based on a laser range finder, including significantly reduced acquisition time. The sensor package used here also allows for considerable expansion of this behavior. For example, complete automation of the building exploration task driven by a mapping algorithm and higher level planner is presently under development.
We explore the potential on-line adjustment of sensory controls for improved object identification and discrimination
in the context of a simulated high resolution camera system carried onboard a maneuverable robotic
platform that can actively choose its observational position and pose. Our early numerical studies suggest the
significant efficacy and enhanced performance achieved by even very simple feedback-driven iteration of the view
in contrast to identification from a fixed pose, uninformed by any active adaptation. Specifically, we contrast the
discriminative performance of the same conventional classification system when informed by: a random glance
at a vehicle; two random glances at a vehicle; or a random glance followed by a guided second look. After
each glance, edge detection algorithms isolate the most salient features of the image and template matching
is performed through the use of the Hausdor↵ distance, comparing the simulated sensed images with reference
images of the vehicles. We present initial simulation statistics that overwhelmingly favor the third scenario.
We conclude with a sketch of our near-future steps in this study that will entail: the incorporation of more
sophisticated image processing and template matching algorithms; more complex discrimination tasks such as
distinguishing between two similar vehicles or vehicles in motion; more realistic models of the observers mobility
including platform dynamics and eventually environmental constraints; and expanding the sensing task beyond
the identification of a specified object selected from a pre-defined library of alternatives.
We document initial experiments with Canid, a freestanding, power-autonomous quadrupedal robot equipped
with a parallel actuated elastic spine. Research into robotic bounding and galloping platforms holds scientific and
engineering interest because it can both probe biological hypotheses regarding bounding and galloping mammals and
also provide the engineering community with a new class of agile, efficient and rapidly-locomoting legged robots. We
detail the design features of Canid that promote our goals of agile operation in a relatively cheap, conventionally
prototyped, commercial off-the-shelf actuated platform. We introduce new measurement methodology aimed at
capturing our robot’s “body energy” during real time operation as a means of quantifying its potential for agile behavior.
Finally, we present joint motor, inertial and motion capture data taken from Canid’s initial leaps into highly energetic
regimes exhibiting large accelerations that illustrate the use of this measure and suggest its future potential as a platform
for developing efficient, stable, hence useful bounding gaits.
For mobile robots, the essential units of actuation, computation, and sensing must be designed to fit within
the body of the robot. Additional capabilities will largely depend upon a given activity, and should be easily
reconfigurable to maximize the diversity of applications and experiments. To address this issue, we introduce a
modular architecture originally developed and tested in the design and implementation of the X-RHex hexapod
that allows the robot to operate as a mobile laboratory on legs. In the present paper we will introduce the
specification, design and very earliest operational data of Canid, an actively driven compliant-spined quadruped
whose completely different morphology and intended dynamical operating point are nevertheless built around
exactly the same "Lab on Legs" actuation, computation, and sensing infrastructure. We will review as well,
more briefly a second RHex variation, the XRL platform, built using the same components.
We discuss the gait generation and control architecture of a bioinspired climbing robot that presently climbs a variety of vertical surfaces, including carpet, cork and a growing range of stucco-like surfaces in the quasi-static regime. The initial version of the robot utilizes a collection of gaits (cyclic feed-forward motion patterns) to locomote over these surfaces, with each gait tuned for a specific surface and set of operating conditions. The need for more flexibility in gait specification (e.g., adjusting number of feet on the ground), more intricate shaping of workspace motions (e.g., shaping the details of the foot attachment and detachment trajectories), and the need to encode gait "transitions" (e.g., tripod to pentapod gait structure) has led us to separate this trajectory generation scheme into the functional composition of a phase assigning transformation of the "clock space" (the six dimensional torus) followed by a map from phase into leg joints that decouples the geometric details of a particular gait. This decomposition also supports the introduction of sensory feedback to allow recovery from unexpected event and to adapt to changing surface geometries.
We review a large multidisciplinary effort to develop a family of autonomous robots capable of rapid, agile maneuvers in and around natural and artificial vertical terrains such as walls, cliffs, caves, trees and rubble. Our robot designs are inspired by (but not direct copies of) biological climbers such as cockroaches, geckos, and squirrels. We are incorporating advanced materials (e.g., synthetic gecko hairs) into these designs and fabricating them using state of the art rapid prototyping techniques (e.g., shape deposition manufacturing) that permit multiple iterations of design and testing with an effective integration path for the novel materials and components. We are developing novel motion control techniques to support dexterous climbing behaviors that are inspired by neuroethological studies of animals and descended from earlier frameworks that have proven analytically tractable and empirically sound. Our near term behavioral targets call for vertical climbing on soft (e.g., bark) or rough surfaces and for ascents on smooth, hard steep inclines (e.g., 60 degree slopes on metal or glass sheets) at one body length per second.
Casual daily observation provides convincing evidence that animals offer a wealth of inspiration for legged machines. However the lessons of animal motor science are largely written in the grammar of materials properties, and their meaning hidden by the complex interaction of multiply layered functional hierarchies. This paper will review some of the lessons of biological running that we have been able to articulate and begin to prescribe rigorously as manifest in the hexapod robot RHex. Although there is a long way to go before our mathematical analysis catches up with the full range of behaviors this remarkable machine exhibits, we are nevertheless able to make increasingly precise statements about certain control principles and the role they may play in RHex's performance. This ongoing research effort serves as a test case to underscore the huge and still largely untapped potential for mining bioinspiration in legged locomotion systems.