In-situ Self-recovery Oriented Contingency Planning Of Autonomous Robots With Chameleon-inspired Perception

With the rapid development of science and technology,autonomous mobile robots are gaining popularities in various fields,especially replacing humans to work in hostile and unknown environments.To improve their abilities adapting to unstructured environments,on one hand,robots need to be equipped with the ability of environment perception like animals.On the other hand,due to the complexity and uncertainty of outdoor environments,robots need to have the ability of contingency planning to cope with emergencies,improving reliability.Consequently,in this paper,in-situ self-recovery oriented contingency planning of wheeled mobile robots(WMRs)with chameleon-inspired perception is studied,mainly including the following aspects:Firstly,to enable mobile robots perceive environments like animals,a multivariate coupling bionic modeling approach is proposed combining BioTRIZ and extension theory,and it is used to realize the design of a chameleon-inspired visual system for a WMR.BioTRIZ is adopted to obtain the proper biological prototype – chameleon's eyes,and the extension conjugate analysis is carried out.A multivariate coupling bionic extension model of chameleon-inspired visual system is built,and the extension hierarchical analysis for multivariate coupling bionic design is implemented.Through the above analysis,the mapping relationship between biological model and bionic model is clarified quantitatively.Based on a 4WD4WS(four-wheel-driving and four-wheel-steering)mobile robot,a chameleon-inspired binocular PTZ(Pan-Tilt-Zoom)visual system is established,and a target search model with chameleon-inspired binocular negative correlation movement(CIBNCM)is built.The similarity between the bionic model and biological model and performance of multivariate coupling bionic design are evaluated,to verify the feasibility of using biological prototype to solve engineering problem.Secondly,inspired by the shifty-behavior mode of chameleon perception,a chameleon-inspired environment perception strategy is presented for WMRs,which can realize the vision-motor coordination of chameleon-inspired visual system and robot-body in the process of target tracking.In the search stage,the visual system uses the CIBNCM and camera short-focus modes to scan the environment in a wide range and high efficiency,and the selective attention algorithm(SAA)is used to achieve the detection of targets.In the tracking stage,the visual system adopts the binocular coordinated movement and camera long-focus modes to achieve the stereo vision,realizing the accurate tracking of targets.The rapid and accurate targeting is realized using the targeting models with the robot-body and neck rotation and zoom,and the particle filter algorithm(PFA)is used for the real-time tracking of targets.Target tracking experiments are implemented in flat and sandy grounds to verify the validity of the proposed approach.Thirdly,the failure mode and effect analysis(FMEA)method and extension analytic hierarchy process(EAHP)are adopted to do risk evaluation for the visual system of WMRs.In terms of the main failure mode of visual perception of WMRs – visual contamination,a field of view(FOV)mosaic perception strategy based on camera movement is presented,and contingency planning based on case-based reasoning(CBR)is deployed to enable robots to perceive environments continuously.A CBR-based contingency planning model of visual contamination is set up,to analyze the contingency planning principle using CBR for visual contamination.In order to get the accurate descriptive parameters of visual contamination cases,an improved contaminant extraction algorithm is proposed,which combines the frame difference and background difference methods based on the Gaussian mixture model.A FOV mosaic perception strategy based on camera movement is presented according to different visual contamination cases,and a perception model with visual contamination is established.The validity of the proposed approach is verified by experiments in flat and sandy grounds based on the conventional planning of target tracking.Finally,the failure mode and effect analysis(FMEA)method and extension analytic hierarchy process(EAHP)are adopted to do risk evaluation for the walking system of WMRs.In view of the main failure mode of movement of WMRs – driving failure,a configuration transformation strategy is proposed,and CBR-based contingency planning is carried out to keep their continuously mobility.A CBR-based contingency planning model of driving failure is established,to discuss how to use CBR for the contingency planning of driving failure.A configuration transformation strategy is proposed according to different driving failure modes,and a movement model with driving failure is set up for 4WD4 WS mobile robots.The quasi static force analysis is deployed,to discuss the force distributions under different configurations and driving failure modes.In order to improve the mobility of mobile robots when driving failure occurs,the traction optimization control is carried out.Experiments are implemented in flat and sandy grounds to verify the effectiveness of the proposed approach based on the conventional planning of target tracking.