WP4 Experimental Study of Motion
Action 4-1: Motion experiments on animals and humans (UTH). During this action accurate measurements of robot motion characteristics during walk and run on uneven terrain will be examined and compared with gait analysis data from individuals with similar anatomical characteristics. Robots’ kinematics will be recorded using our 10 camera Vicon motion analysis. Multiple markers will be positioned on each segment to obtain realistic 3D motions. Moreover ground reaction forces during robots’ gait will be measured. The outcomes of this action will help NTUA and FORTH team to optimize the various motion control parameters of robots’ system and computational models.
Action 4-2: Experiments on algorithms and control methods for four-legged robots (NTUA). After the quadruped robot is developed, the algorithms, which have been tested computationally, will be applied on the robot. It is evident that the theoretical model and the actual robot will not be an exact match, since the actual robot is much more complex and there are many undefined parameters which affect it, which are not included in the Matlab model. Therefore, all three parts, the control algorithm, the theoretical model and the developed robot, must be changed so that the actual robot will move as it is supposed to, according to the theoretical model. This will require a large amount of effort, and it is possible that different approaches in both the control and the mechanical part of the robot will need extensive changes to accomplish the goal. During the whole procedure, UTH will be aiding NTUA with information on how nature has overcome the difficulties that will be encountered.
Action 4-3: Experiments on multi-legged robots (FORTH). A comprehensive range of experiments with multi-legged prototypes will be performed, focusing on motion over unstructured and/or uneven terrain, to assess the efficacy of gait generation strategies. The results of these will be employed towards optimizing the various motion control parameters, and refining the computational models. Towards the latter goal, special testbeds and experimental procedures will be set up for characterizing the interaction of the robot with the locomotion substrate. The potential of employing motion capture equipment at UTH and NTUA, for obtaining accurate robot motion measurements will be explored. The experimental assessment of the sensor-based closed loop behaviours, (WP2), will be pursued. Action 4-4: Experiments on biped robots (FORTH). The availability of a humanoid platform with the necessary characteristics (FORTH currently owns a FUJITSU HOAP-3 humanoid robot), will enable the implementation of the locomotion control strategies designed, and studied in simulation, in WP2. Experimentation will involve the calibration of the robot, the characterization of its sensory apparatus and the implementation of a set of basic behaviours, on which further developments will be based. The experiments will include different types of locomotion behaviours (e.g., walking, running or jumping), as well as different types of environments (e.g., inclined surfaces, stairs), focusing on efficiency and robustness. The obtained results will be systematically compared to the human data obtained from UTH, and appropriate modifications of the control strategies will be investigated.