We analyze the the lateral rolling type of locomotion, present path planning of shape transition, and develop a 3-dimensional snake-like robot. First we propose one of a lateral rolling type of locomotion “Twisting mode” and analyze the principle of the locomotion based on statics. Next, the motion planning using Genetic Algorithm for transition of locomotion modes is explained and simulation results for the GA research are shown. Finally, we explain the designed prototype system and experimental results are shown.
This paper describes a real-time human tracking system by audio-visual integrtation for the humanoid SIG. An essential idea for real-time and robust tracking is hierarchical integration of multi-modal information. The system creates three kinds of streams - auditory, visual and associated streams. An auditory stream with sound source direction is formed as temporal series of events from audition module which localizes multiple sound sources and cancels motor noise from a pair of microphones. A visual stream with a face ID and its 3D-position is formed as temporal series of events from vision module by combining face detection, face identification and face localization by stereo vision. Auditory and visual streams are associated into an associated stream, a higher level representation according to their proximity. Because the associated stream disambiguates parcially missing information in auditory or visual streams, “focus-of-attention” control of SIG works well enough to robust human tracking. These processes are executed in real-time with the delay of 200 msec using off-the-shelf PCs distributed via TCP/IP. As a result, robust human tracking is attained even when the person is visually occluded and simultaneous speeches occur.
Reinforcement learning is very effective for robot learning. Because it does not need priori knowledge and has higher capability of reactive and adaptive behaviors. In our previous works, we proposed new reinforce learning algorithm: “Q-learning with Dynamic Structuring of Exploration Space Based on Genetic Algorithm (QDSEGA) ”. It is designed for complicated systems with large action-state space like a robot with many redundant degrees of freedom. However the application of QDSEGA is restricted to static systems. We extend the layered structure of QDSESA so that it could be applicable to the dynamical system. A snake-like robot has many redundant degrees of freedom and the dynamics of the system are very important to complete the locomotion task. For this task, application of usual reinforcement learning is difficult. In this paper, we extend layered structure of QDSEGA for applying real robot. We apply it to acquiring of locomotion pattern of the snake-like robot and demonstrate the validity of QDSEGA with the extended layered structure by simulation and experiment.
This paper addresses a feedback control strategy for 3D free-flying linkage robots with non-zero angular momentum, in which the robots' orientation and configuration can be reached to desired states in an arbitrary amount of time. Mainly since its constraint is non-Chaplygin, their motion is very complicated to grasp intuitively. We first analyze the dynamic mechanism of twisting somersault motion and then design the controllers to make the systems perform this kind of motion. The proposed controllers characteristically utilize the feedback against the system's limit state. Several simulation results show their readily reachability and robustness with initial state error. The controller is also applicable for plannar systems and spinning satellites in space.
Typical human-based biped robot has 12-DOF for its legs, but there may be another approach not to take a model on human, but to design function oriented. To design a practical light-weight biped, reducing a number of DOF is one of the most efficient method. This paper discusses minimum number of DOF for various cases, where 4, 7, 5 is the resultant minimum number of DOF for horizontally composed terrain, uneven terrain with normal leg exchange sequence, uneven terrain with partial leg exchange sequence, respectively. This paper also shows practical designs of bipeds with 5, 8, and 6-DOF. Trial manufactured machines based on these discussions could show good performances on stairs and uneven terrain.
Humans can easily maneuver some types of nonholonomic systems, e.g. wheeled vehicles, while other types, e.g. space robots, are difficult to handle intuitively. We propose a human interface to simplify the operation of “difficult” nonholonomic systems, which utilizes the human ability to maneuver “easy” systems. The difficult real system is converted into an easy virtual system using coordinate and input transformation. The input from the human operator to the virtual system is converted into input to the real system, while the state of the real system is converted into that of the virtual system which is displayed to the operator. The operator can then steer the real system feeling as if maneuvering the virtual system. Our experiments show that the operating performance is improved by this method.
A new method to form feedforward inputs for underwater robot manipulators is proposed in this paper. In the proposed method, hydrodynamic torques such as added mass, drag and buoyancy in dynamics of underwater robots are obtained by time scale transformation and iterative learning control. The advantage of the proposed method is not to use parameter estimation of the dynamics. The results of the fundamental experiments with a single-link arm demonstrate validity of the iterative learning control in water and acquisition of the accurate hydrodynamic torques by time scale transformation for control.
We have been trying to induce a quadruped robot to walk with medium walking speed on irregular terrain based on biological concepts. In this paper, we define adaptive walking based on biological concepts as “coupled-dynamics-based motion generation”, in which a neural system and a mechanical system are coupled and generate motion by interacting with the environment emergently and adaptively. We design the mechanical system and the neural system consisting of a central pattern generator (CPG), responses and reflexes. A CPG receives sensory input and changes the period of its own active phase as responses. Especially, we propose that the rolling motion should be input to CPGs in order to synchronize the pitching and rolling motions. PD-controller at joints as the stretch reflex constructs the virtual spring-damper system as the visco-elasticity model of a muscle. The desired angle and P-gain of each joint in the virtual spring-damper system is switched based on the phase signal of the CPG. CPGs, the motion of the virtual spring-damper system of each leg and the rolling motion of the body are mutually entrained through the rolling motion feedback to CPGs, and can generate adaptive walking on irregular terrain. The mutually entrained system closely couples the representative indices such as walking speed, gaits, stability and energy consumption . We report our experimental results of dynamic walking on terrains of medium degrees of irregularity in order to verify the effectiveness of the designed neuro-mechanical system.