Research Of Key Technologies On Security Robot Systems In Residential Districts

At present, as population density is growing, residential districts are increasing in scale.Theft cases also occur frequently in residential districts. The difficulty and securitypersonnel’s dereliction of duty make it hard to eradicate these cases, although a variety ofcountermeasures have been taken by each residential district. As a result, developing anddeploying security robots into residential districts will be a very promising research topic.This research can let security robots complete tasks such as daily patrol and thief catching inresidential districts. Under such social background, the fundamental purpose of this topic is tosolve problems such as multistage hierarchical motion planning, and collaboration of securitypersonnel and multiple security robots in chasing and catching thieves. The expectation ofthis research is to prevent thieves from thieving at the most extent, and control the thievesafter their crime as soon as possible, which will let them have no any opportunity to commit acrime.This paper not only presents the fundamental idea of security robot’s system architecture,but also discusses key questions such as the overall structure, detail content of each majorpart, implementation methods, and final results.First of all, this paper splits and merges each region of the residential districts on themultistage hierarchy, and gives a definition for the samples such as buildings, roads, androckeries, at the same time this paper achieves environment modeling of the whole residentialdistricts on the semantic and space level using knowledge available.Second, this paper researches the multistage hierarchical motion planning in furtherdepth and detailed, and applies it to the security robot motion planning. According to theassignment requirements, this paper design the next sequence or subtasks with the multistagehierarchical motion planning in order to achieve the assignment-level planning of securityrobots; Using the subordinate relations, as well as connecting and adjacency relations of theresidential districts to achieve the semantic-level motion planning of the security robots. Baseon assignment-level and sematic-level motion planning, this paper further achieves thespace-level and action parameter-level motion planning of security robots.Moreover, this paper analyzes the problems of multi-robot collaboration in thief catching by further exploring sub questions of robot learning, multi-robot collaboration, and optimaltask allocation mechanisms, so as to solve it.Finally, this paper describes the function modules and experimental system design indetail, and gives the actual running result of each module, and achieves simulation on system.The end of this paper points out some weaknesses that are still exist in the system, and alsodirections of future work.