This thesis presents methods for implementing robust hexpod locomotion on an autonomous robot with many sensors and actuators. The controller is based on the Subsumption Architecture and is fully distributed over approximately 1500 simple, concurrent processes. The robot, Hannibal, weighs approximately 6 pounds and is equipped with over 100 physical sensors, 19 degrees of freedom, and 8 on board computers. We investigate the following topics in depth: distributed control of a complex robot, insect-inspired locomotion control for gait generation and rough terrain mobility, and fault tolerance. The controller was implemented, debugged, and tested on Hannibal. Through a series of experiments, we examined Hannibal's gait generation, rough terrain locomotion, and fault tolerance performance. These results demonstrate that Hannibal exhibits robust, flexible, real-time locomotion over a variety of terrain and tolerates a multitude of hardware failures.

No description available.

RHex is an autonomous hexapedal robot capable of locomotion on rough terrain. Up to now, most modelling and simulation efforts on RHex were based on the linear leg assumption. These models disregarded what might be seen as the most characteristic feature of the latest iterations of this robot: the half circular legs. This thesis focuses on developing a more realistic model for this specially shaped compliant leg and studying its effects on the kinematics and dynamics of the resulting platform. One important consequence of the half circular compliant leg is the resulting rolling motion. Due to rolling, the rest length of the leg changes and the leg-ground contact point moves. Another consequence is the varying stiffness of the legs due to the changing rest length. These effect the resulting behaviour of any platform using these legs. In the first part of the thesis we are studying the effects of the half circular leg morphology on the kinematics of RHex using a simple planar model. The rest of the studies within the scope of this thesis focuses on the effect of the half circular compliant legs on the dynamics of a single legged hopping platform with a point mass. The formulation derived in this work is successfully integrated in a readily working but rather simple model of a single legged hopping system. We replace the equations of the straight leg in this model by the equations of the half circular compliant leg. Realistic results are obtained in the simulations and these results are compared to those obtained by the simpler constant stiffness straight leg model. This more realistic leg model brings us the opportunity to further study the effects of this leg morphology, in particular the positive effects of the resulting rolling motion on platform stability.

During the uniform locomotion of legged robots with compliant legs, the body of the robot exhibits quasi-periodic oscillations that have a disturbing eect on dierent onboard sensors. Of particular interest is the camera sensor which suers from image degradation in the form of motion-blur as a result of this camera motion. The eect of angular disturbances on the camera are pronounced due to the perspective projection property of the camera. The thesis focuses on the particular problem of legged robots exhibiting angular body motions and attempts to analyze and overcome the resulting disturbances on a camera carrying platform (head). Although the full problem is in 3D with three independent axes of rotation, a planar analysis provides signicant insight into the problem and is the approach taken in the thesis. A carefully modeled planar version of an actual camera platform with realistic mechanical and actuator selections is presented. Passive (ltering) and active (controller) approaches are discussed to compensate/cancel motion generated disturbances. We consider and comparatively evaluate PID and LQR based active control. Since PID has the limitation of controlling only one output, PID-PID control is considered to iv control two states of the model. Due to its state-space formulation and the capability of controlling an arbitrary number of states, LQR is considered. In addition to standard reference signals, Gyroscope measured disturbance signals are collected from the actual robot platform to analyze the bandwidth and test the performance of the controllers. Inverted pendulum control performance is evaluated both on a Matlab-Simulink as well as a precise electro-mechanical test setup. Since construction of the planar head test setup is in progress, tests are conducted on simulation.

A dynamically dexterous legged robot platform generates specific acoustic signals during the interaction with the ground and the environment. These acoustic signals are expected to contain rich information that is related to the interaction surface as a function of the position of the legs and the overall contact process mixed with the actuator sounds that initiate the movement. As the robot platform walks or runs in any environment, this convolved acoustic signal created can be processed and analyzed in real time operation and the interaction surface can be identified. Such an utilization of acoustic data can be possible for various indoor and outdoor surfaces and with this can be useful in adjusting gait parameters that play an essential role in dynamic dexterity. In this work, surface type identification is achieved with using the several popular signal processing and pattern classification methods not on the robot platform but off-line. The performances of the selected features and the algorithms are evaluated for the collected data sets and these outputs are compared with the expectations. Depending on the off-line training and experiment results, the applicability of the study to an embedded robot platform as a future application is found quite feasible and the surface type as an input to the robot sensing is expected to improve the mobility of the robot in both indoor and outdoor environment.

No description available.

No description available.

Recent interest in using passive dynamic walkers (PDWs) for gait rehabilitation studies has presented a need for a robust, easily built mechanism. Unfortunately, these passive robots are hypersensitive to many variables outside of the usual design considerations that are studied when constructing them. By accentuating previous failures instead of suppressing them, this thesis presents a number of problematic situations commonly experienced when testing and tuning a PDW. Further, through a complete design of a 4-legged PDW with knees, simple design axioms brought about by myself and others are put into a practical context and applied directly to design. This thesis aspires to present a systematic design process, and highlight how a computational model can be used with both hand calculations and CAD packages. Using the insight from those researchers before me, I strive to further their designs and present relevant information in a design compendium that makes it more useful to those who have an application for the device. This thesis resulted in two novel designs for a PDW. First, a changing radius foot was developed to increase knee flexion upon toe off. The decrease in radius increases joint angular velocity resulting in ramp up. Further investigation of these feet could result in more stable and efficient walking patterns. The other design brought to attention is the planar crossbar mechanism for coupling the inner and outer legs. The crossbar provides a rigid coupling without changing the rotational inertia between the coupled pair about the hip axis.

Unstable legged robots fall over without active stabilization, typically by repositioning the feet to maintain/regain stability of balance. This dissertation concerns the development of a Single-Track Three Legged Mobile Robot (ST3LMR) and control system. A proof-of-concept was demonstrated through digital simulation and experimentation with physical prototypes. The ST3LMR comprises a body and three articulated legs arranged in a narrow profile, one behind the other, to walk and maneuver along narrow trails and paths. The ST3LMR walks by placing successive footfalls in a generally single-track or in-line fashion. It achieves the form and function of a motorcycle but with the added benefit of legs and robotic control. That is, the feet are stationary with respect to footholds during the support period, thus eliminating the drawback of wheels, which require continuous support (especially when used in rugged terrain). By always having at least two feet on the ground, the ST3LMR is inherently stable in the pitch axis (in the forward/backward direction of motion), which allows for decoupling stability of balance control to only the roll axis (in the left/right direction).Suggested by recent developments in high-performance computing, walking robot locomotion and stabilization is considered from a new perspective, that of the Monte Carlo (MC) method. A high-speed MC simulation is used in a model-predictive control system to determine footholds that provide stability of balance. Stability of balance, maneuverability, and control is demonstrated through experimental results from physical prototypes and a simple digital simulation of an impulse response, avoidance maneuver, and leaning-into-the-turn maneuver.

We present a novel control strategy for running which is robust to disturbances, and makes excellent use of passive dynamics for energy economy. The motivation for our control strategy is based on observations of animals, which are able to economically walk and run over varying terrain and ground dynamics. It is well-known that steady-state animal running can be approximated by spring-mass models, but these passive dynamic models describe only steady-state running and are sensitive to disturbances that animals can accommodate. While animals rely on their passive dynamics for energy economy, they also incorporate active control for disturbance rejection. The same approach can be used for spring-mass walking and running, but an active controller is needed that interferes minimally with the passive dynamics of the system. We demonstrate, in simulation, how force control combined with a leg spring stiffness tuned for the desired hopping frequency provides robustness to disturbances on a model for robot hopping, while maintaining the energy economy of a completely passive system during steady-state operation. Our strategy is promising for robotics applications, because there is a clear distinction between the passive dynamic behavior of the model and the active controller, it does not require sensing of the environment, and it is based on a sound theoretical background that is compatible with existing high-level controllers for ideal spring-mass models. / Graduation date: 2012