BigDog is a four legged robot with exceptional rough-terrain mobility. In this paper, we equip BigDog with a laser scanner, stereo vision system, and perception and navigation algorithms. Using these sensors and algorithms, BigDog performs autonomous navigation to goal positions in unstructured forest environments. The robot perceives obstacles, such as trees, boulders, and ground features, and steers to avoid them on its way to the goal. We describe the hardware and software implementation of the navigation system and summarize performance. During field tests in unstructured wooded terrain, BigDog reached its goal position 23 of 26 runs and traveled over 130 meters at a time without operator involvement.

SDR-4X is a small biped entertainment robot with expanded capabilities for a variety of motion performances. We developed the SDR motion creating system and the whole body adaptive motion control for SDR. This software system makes it easy to produce creative motion performances that include stable biped walking and dance steps synchronized to music. The motion creating system is composed of two parts: the motion editor used to edit upper body motion and whole body motion, and the foot trajectory editor used to create stable lower body motion. Once the upper and lower body motions are created, the system allows the motion designer to inspection the whole body motion before applying it to SDR-4X. The process of creating a fast-paced dance performance is introduced as an attractive application of this software system.

Summary form only given. We are interested in physical robots that move in the physical world and in simulated creatures that move in the virtual world. We have built a series of physical robots that include one-legged hoppers, biped runners, a quadruped, and two kangaroo-like robots. These robots use simple control systems to run with several gaits, go fast, traverse simple paths, jump obstacles, climb a simplified stairway, and perform rudimentary gymnastic maneuvers. Although no one robot performs all these task, they all use a common set of principles for their control. We think that active balance, springy legs, and dynamic control are the features responsible for whatever "lifelike" behavior these machines have. Recently we have begun to use ideas about robot control to control animated creatures. The idea is to give these creatures knowledge and control of their body motion so the animator can act more like a director, working at a relatively high level. If the creature knows how to move in response to task-level instructions, then the director can concentrate on creating the overall action and story.

The somersault is a maneuver in which a performer jumps into the air and rotates through 360 degrees before landing on the ground. The somersault is challenging to execute because the performer has limited control over the body while airborne. Despite conservation of momentum con- straints, limited control of such a maneuver during flight is possible. We experiment with the use of an in flight feedback strategy to control the somer- sault landing of a biped robot. In flight, the robot manipulates its rotation rate by tucking its legs to control the landing attitude. Results from somer- sault experiments in the laboratory provide initial verification of the control strategy.

An articulated leg with a hoof is studied in the context of a planar one-legged running machine, the monopod. The monopod runs forward, backward, and in place at relatively low speed. One task for controlling this design is to keep the hoof from rolling. Rolling may be caused by hip torque from the body attitude control and the inertial loading from the acceleration of the monopod during the stance phase. It is shown that coordinating the hip torque with the downward force at the hoof does not control hoof rolling due to excessive pitch disturbance. The method of limiting the inertial loading successfully prevents rolling of the hoof.>

This paper is about the use of control algorithms to animate dynamic legged locomotion. Control could free the animator from specifying the details of joint and limb motion while producing both physically realistic and natural looking results. We implemented computer animations of a biped robot, a quadruped robot, and a kangaroo. Each creature was modeled as a linked set of rigid bodies with compliant actuators at its joints. Control algorithms regulated the running speed, organized use of the legs, and maintained balance. All motions were generated by numerically integrating equations of motion derived from the physical models. The resulting behavior included running at various speeds, traveling with several gaits (run, trot, bound, gallop, and hop), jumping, and traversing simple paths. Whereas the use of control permitted a variety of physically realistic animated behavior to be generated with limited human intervention, the process of designing the control algorithms was not automated: the algorithms were "tweaked" and adjusted for each new creature.

The task of controlling step length in the context of a dynamic biped robot that actively balances itself as it runs is discussed. Three methods for controlling step length, each of which adjusts a different parameter of the running cycle, are discussed. The adjusted parameters are forward running speed, running height, and duration of ground contact. All three control methods are successful in manipulating step length in laboratory experiments, but the method that adjusted forward speed provided the widest range of step lengths with accurate control of step length. The three methods for controlling step length manipulated the dynamics of the system so the feet could be placed on the available footholds without disturbing the system's balance. An alternative approach which ignores balance for a single step, placing the foot directly on the desired foothold, and recovering balance later is described.>

Even and odd functions of time - symmetric functions - have been used to simplify the control of running robots and have been observed in the body and leg motion of running animals. This paper explores the relationship between symmetric motion of the body and symmetric use of the legs in three planar models. The algebra of symmetric functions is applied to the equations of motion to show that symmetric leg actuation is required for symmetric body motion in the single-support case, but not in the double-support case unless additional constraints are imposed.

Simple locomotion algorithms provide balance for machines that run on one leg. The generalization of these one-leg algorithms for control of machines with several legs is explored. The generalization is quite simple when muitilegged systems run with gaits that use the support legs one at a time. For these gaits the one-leg algorithms can be used to control multilegged running. The concept of a virtual leg is introduced to further extend the approach to gaits that use the legs in pairs, such as the trot, the pace, and the bound. These quadruped running gaits map into gaits that use one virtual leg for support at a time, for which the one-leg algorithms can provide control. This approach was used in laboratory experiments to control a quadruped machine that runs with a trotting gait.

This paper describes a new VLSI tactile array sensor that was designed for use on the fingers of a robot manipulator. The photolithographic methods used in fabricating the microelectronics were also used to build an array of mechanical structures on top of an integrated circuit. When covered with a suitable elastic material, the mechanical structures, notch-shaped windows in the protective overglass, convert applied pressure directly into a set of switch-like closures. The data obtained from these switch closures are multiplexed by logic in the circuit before transmission to a host, thereby reducing the wiring problem associated with discrete array sensors. A prototype with 48 cells arranged in a 6 by 8 array is currently being built.

A two-dimensional one-legged hopping machine is modeled and simulated in order to better understand legged systems that hop and run. The analysis is focused on balance, dynamic stability, and resonant oscillation for the planar case. A springy leg with nonzero mass, a simple body, and an actuated hinge-type hip are incorporated in the model. Control of the model is decomposed into a vertical hopping part, a horizontal velocity part, and a body attitude part. Estimates of total system energy are used in regulating hopping height in order to initiate hopping, to maintain level hopping, to change from one hopping height to another, and to terminate hopping. Balance and control of forward velocity are explored with three algorithms. The feasibility of decomposing the control of running into a height control part, a forward velocity control part, and an attitude control part is verified by simulations.

A method for control of dynamic systems which uses a large table of precomputed data is described. The method was developed in the context of controlling legged systems that balance as they run. The tabular approach takes advantage of the very regular, cyclic character of legged behaviour in order to partition the system state variables into two sets: one set that varies in a relatively fixed way on each cycle, and another set that varies freely from cycle to cycle. Only this second subset of the state variables determines the size of the table. Stored data are computed by numerically simulating a dynamic model of the legged system as it progresses through the stance portion of the running cycle. Repeated simulations are used to characterize the system for different landing conditions. Polynomial surfaces are used to approximate the tabular date. The feasibility of the methods is shown by simulations.

The feasibility of applying robot manipulators to complex handling and assembly problems in indus try and space depends on the ability to control manipulators precisely, even when there are uncertainties and variations in the environment. This can be done by giving a manipulator hand sensors that provide information about the progress of interactions with the environment. A method is presented for controlling both the position of a manipulator and the contact forces generated at the hand. This hybrid controller has the feature of permitting trajectories to be specified in a convenient Cartesian coordinate system, while offering advantages over the existing approximate "free joint" methods. An experimental implementation of the controller was developed and used to control two joints of a modified Scheinman manipulator equipped with a wrist mounted force sensor. Data from this implementation are presented along with the results from a simple simulation.