THE PUBLISHING HOUSE PROCEEDINGS OF THE ROMANIAN ACADEMY, Series A, OF THE ROMANIAN ACADEMY Volume 11, Number 1/2010, pp. 75–82 AN APPROACH TO WALKING ROBOTS PLANNING AND CONTROL Sergiu Boris CONONOVICI*, Adrian CURAJ** * Institute of Solid Mechanics of Romanian Academy, C-tim Mille 15, Bucharest 1, +40213126736, luigiv@arexim.ro ** “Politehnica” University of Bucharest, Splaiul Independentei 125, Bucharest 5, Romania Vehicles which operate more or less autonomously are considered belonging to the robot’s family. The legged vehicles have the potential advantage to be able to operate over irregular terrain. The planning and control of these robots is considered hierarchically organized on two levels. At the high level, the robot is piloted in the task world. At the local level – operative level – the locomotion function is performed. In this paper, walking orthogonal hexapods, for moving along uneven terrain are the object of study. By kinematic analysis the direct and inverse geometrical models and a differential model are developed. The accommodation to terrain using force sensing is approached too. This approach is useful to write the control algorithms of the analyzed robots or of other structures by analogy. Key words: Robot control; Walking robots; Mathematic modelling. 1. INTRODUCTION Vehicles which operate more or less autonomously are considered as belonging to the robot’s family. Betwixt these, the legged vehicles have the potential advantage to be able to operate over irregular terrain. Walking robots (WR), in comparison with industrial robots which are dextrous arms (6 or more D.O.F.) able to ensure the precise positioning of the end-effector, comprise a body and legs, each kinematic chains with no more than 3 D.O.F. which support the body and ensure the gait, the locomotion mechanism. Terrain – adaptive walkers have the ability to individually place their feet at desired position on the terrain [1, 2]. The legs can be in supporting state when the legs are in contact with the ground, and the body is moving or not, with respect to the footholds, or in stepping or transfer state when the leg is detached and has a relative movement to the robot’s body, somehow looking for a place for the new foothold [3, 9, 10]. Fig. 1 – Walking vehicle with weaving gait. With orthogonal walking robots, the body movement is carried out in a plane with a group of actuators, and the vertical movement is performed by another group of actuators [6]. The three-degree-of-freedom orthogonal legs decouple horizontal and vertical motion, leading to simplified planning and control. The legs are kinematic chains with driving (propulsive) joints revolute (R) or prismatic (T), RTT or RRT, with the supporting element pedestal with prismatic joint; the supporting element is always vertical to the foothold [4, 8]. In this paper two orthogonal-legged walker configurations are tackled. The first is termed a weaving walker (Fig. 1) because recovering legs thread through supporting legs.