本文提出以包容式框架結(jié)構(gòu)為基礎(chǔ), 將計劃作為一種行為, 與應(yīng)激式行為并行處理的思想. 與之對應(yīng)的是將傳感器檢測區(qū)域分為避障區(qū)和規(guī)劃區(qū), 如果在避障區(qū)有障礙物, 則由應(yīng)激避障模塊采取相應(yīng)的避障行為; 如果在規(guī)劃區(qū)有障礙物, 則由規(guī)劃模塊產(chǎn)生相應(yīng)的輸出. 由于規(guī)劃模塊最終輸出子目標(biāo), 因此使此結(jié)構(gòu)的規(guī)劃模塊具有很好的擴展性. 另外, 在處理應(yīng)激避障行為和直接規(guī)劃行為時, 分別提出新的基于傳感器的避障和規(guī)劃方法, 通過機器人路徑規(guī)劃的仿真證明, 這些方法是簡潔、有效的.
關(guān)鍵詞: 混合式結(jié)構(gòu); 避障; 基于傳感器的旋轉(zhuǎn)法; 規(guī)劃; 機器人
　Abstract: In th is paper, w e describe a hybrid cont ro l arch itecture based on Subsum ing Cont ro lA rch itecture. The p lanning is regarded as a behavio r, and p rocessed synch ronist ically w ith the react ive behavio rs. Co rrespondingly, w e part it ion the detected field into an obstacle2avo idance field and a p lanning field. If an obstacle has been detected in the obstacle2field avo idance , the react ive obstacle2avo idancemodule p rocesses it, and if an obstacle has been de2 tected in p lanning field, the p lanning module p rocess it. The simulat ions of robo t path p lanning p rove that the method can fuse the react ive behavio r and the p lanning behavio r perfect ly. In addit ion, w e have given a new reac2 t ive obstacle2avo idance method and a new p lanningmethod based on senso rs. Thesemethods are p roved simp le and effect ive.
Keywords: hybrid arch itecture, avo iding obstacle, RBS, p lanning, robo t