The program includes analytical, simulation and experimental thrusts. The analytical work will focus on spatial dynamic modelling and passive dynamics analysis of highly articulated, variable compliance quadrupeds and on robust nonlinear control methods for achieving traversing speeds in known but variable environments, with stability and efficiency. A simulation environment will provide a virtual test bed for alternative algorithm and design testing. The experimental work will focus on designing and developing a novel quadruped robotic system, capable of demonstrating the validity of the analytical and simulation results. The proposed research will build upon our previous experience with both nonlinear model-based control of quadrupeds and designing and building such robots in-house.
The main program objectives include:
- Computational test bed. We will develop a computational test bed to automated production of equations of motion in analytical and numerical form using Lagrangian dynamics, the Mathematica s/w package, and the ADAMS s/w package.
- Flexible multi-joint body. We hypothesize that the addition of a transverse rotational degree of freedom to the robot’s main body will allow achieving higher speeds, and reduce the number of leg joints required. To investigate this issue, the computational test bed will be employed, along with model-based nonlinear control, and observations from nature. The results will be used in the design of the multi-jointed quadruped.
- Addition of a tail or momentum devices. We intend to study if using an artificial tail or momentum storage devices (reaction wheels), can enhance stability and capabilities for rapid change of direction
- Multi-joint, variable compliance legs. The introduction of variable leg compliance, especially as a function of ground stiffness and robot dynamics, is a very intriguing research issue that can be achieved by design or control methods, and studied both in simulation and experimentally, at the single leg level.
- Performance criteria. A number of important questions regarding the design and required dofs for quadrupeds remain unanswered. These will be studied using the energetic cost of transport (COT).
- Gait transitions. To achieve maximum speed, or to turn, the capability gait changes must exist. Analysing gait characteristics and planning the transient commands under torque and stability constraints will yield the required transition strategy.
- Novel control algorithms, stability and robustness. The objective is to design robust controllers that guarantee system stability in the presence of small disturbances, and that allow setting desired goals, such as robot speed or apex height. Stability criteria, such as the force-angle criterion, will be employed. Control algorithms will be developed based on the passive dynamics of a quadruped with articulated body and multi-joint legs. Tools from the areas of nonlinear dynamics (nonlinear oscillators, limit cycle analysis, Poincaré maps, etc.) and nonlinear control (underactuated, hybrid, backstepping, and robust model-based control, etc.) will be employed.
- Scaling. Fundamental questions include how energy, efficiency, speed and range scale with size.
- High-speed inter-communication system. A high-speed inter-communication system is needed allowing real-time communication with all actuator/ sensor clients, ensuring a hard-realtime exchange of information.
- Realization of a quadruped robot. A quadruped with multi-jointed legs, articulated body and a novel actuation system, performing dynamic stable motion at high speeds through uneven terrains, will be designed and built. This effort will be assisted by the availability of new high-torque brushless motor technologies, miniaturization of driving and sensing electronics, and with the development of a new high-speed inter-communication system for adding actuators and sensors using Ethernet and the IEEE1588 protocol. The robot will be built in-house at the CSL facilities.