M.Sc Thesis

M.Sc StudentTemtsin Sharon
SubjectDecision Making Algorithms for Safe Robotic Disassembling
of Randomly Piled Objects
DepartmentDepartment of Civil and Environmental Engineering
Supervisor ASSOCIATE PROF. Amir Degani
Full Thesis textFull thesis text - English Version


Autonomous capabilities for manipulating randomly piled objects may enhance current methods of path planning and open a new field of development for mobile manipulation and Urban Search and Rescue (USAR) robotics. Pile disassembling on USAR sites often requires taking into consideration the pile disturbance in order to keep the safety of the survivors, the USAR teams, or the safety of the site’s environment.  Autonomous disassembling of randomly piled objects with minimum disturbance includes three fundamental challenges: object segmentation, decision-making and motion planning. This work introduces decision-making challenges for achieving autonomous manipulation capabilities in pile disassembling process. Four decision-making algorithms including three novel algorithms are proposed for selection of a single object to remove from a pile: the Volume Height Algorithm (VHA), the Filtered Volume Height Algorithm (FVHA), the Kinematic Based Algorithm (KBA) and the Filtered Volume Height Kinematic Based Algorithm (FVHKBA). The VHA removes the object with the highest potential energy. The FVHA dismantles the object with the highest potential energy that does not support any other object from below. The KBA decides to remove the least encapsulated object by determining a movability rank for each object according to the degree of its encapsulation and picking the object with the highest rank. Using the contact vectors of the examined object, it is possible to obtain the motions that will not violate the object’s unilateral contact constraints. The movability rank of the object is proportional to the union of all such motions. The FVHKBA disassembles the less encapsulated object with the highest potential energy that doesn't support any other object from below. All algorithms were tested and analyzed in simulation on piles composed of identical blocks and different blocks under full and partial knowledge conditions. FVHA, a combination of VHA and KBA strategies, shows the most impressive performances regardless of knowledge conditions. This work introduces an autonomous manipulation challenges the solution of which will affect the field of manipulation - particularly USAR. In addition, an experimental evaluation of four algorithms targeted at potential USAR constructions may enhance the field of mobile manipulation and contribute to robot integration in USAR manipulation tasks.