Research On Motion Control Of Amphibious Robot Based On CPG Neural Network

With the ever-accelerating pace of mankind exploration of the unknown world, the demand for mobile robots with high mobility under complicated surroundings is increasingly popular. The amphibious robots have attracted more and more attention from researchers for its strong adaptability on land, water and transition terrain environment and broad prospect of application and major scientific research value in the area of military detection, earthquake relief, space navigation and medical services, meanwhile, it also brings great challenge to researchers for the design of propulsion mechanism and control algorithm.Based on the reference of the propulsion mode of land arc legs and underwater fluctuations, we design an amphibious robot with variable stiffness leg of single degree of freedom, propelled by the curved foot on land, the fluctuation in water and the combination of the two motions in transition terrain. The main research contents and results of this thesis are as follows:Whole system design of robot AmphiHex-Ⅱ. Summarize the experience of forefathers to design a structure of variable stiffness leg which integrates the advantages of curved leg and webbed leg, and design the driver module of single degree of freedom and waterproof case aimed at the unique structure. Plan the block diagram of hardware system and modular design thought of software system.Building the control system based on CPG neural network. Select the appropriate neuron model direct at the motion characteristics of AmphiHex-Ⅱ to establish CPG neural network through the specific topological structure, and debug the related parameters to generate self-excited stable oscillation signals.Planning of movement gait based on CPG neural network. Project the various movement gaits in diverse terrains, including land, water and land-water transition terrain, etc. Establish the correspondence between the control signal based on CPG and various movement gaits and formulate suitable gait control strategy.Motion performance test on robot AmphiHex-Ⅱ. Carry out abundant motion performance experiment in kinds of environment, such as hard ground, sand terrain, grass, slope, steps, water and land-water transition terrain. When walking on hard ground, the average velocity of fast tripod gait is about 0.36m/s (0.7 times of body length), while the average velocity of fast hexapod gait is about 0.26m/s (0.5 times of body length). When swimming in water, the velocity of propelling with 30° swing amplitude and 2 hz swing frequency is about O.11m/s (0.22 times of body length).The robot can climb the step obstacles with the height of 180mm. Demonstrate the climbing gait is more stable than tripod gait when climb a big slope with the dip angle of 25°. Test the effect of turning in various terrains. Realize the diverse maneuvering motions underwater and the landing, diving motion in land-water transition terrains.