Research On Some Frontier Technologies Towards Flexible Service Robot

With the rapid development of robotics, robots have started to go out of the pro-duction line and come to our homes and to serve the people in the future. As a typical nonlinear complex systems and comprehensive integrated system,the development of a service robot involving a wide range of multidisciplinary technologies and faces many big challenges.How to improve the performance of service robots effectively to meet the needs of service tasks becomes a important challenge.This dissertation is based on the Kejia service robot project, which I participated in during my study for my doctorate. The dissertation introduces the design of ser-vice robots, control, planning and other aspects of the content including the frontier of robotics research-soft robotics. In this dissertation, Firstly I describes the current de-velopment of service robots and discusses the reasons why robotic now is getting more and more attention in social circles. Then the characteristics of robots in different coun-tries and some well-known high-performance service robot are listed. The dissertation also analyzed the trend of the current robotics technology based on a market view. After that, the main contents can be divided to three parts:The first part takes Kejia service robot for an example and describes the design of the hardware parts, including the mechanical design, exterior design, dynamic design, control system design and so on.The hardware system is the basis for generating robot behavior, which has a significant influence on the performance of the robot. A Good de-sign of hardware system should consider the operation range and operation type of the robot and place suitable DOFs to enhance its capabilities and reduce the difficulties of controlling such as kinematics computing. A robot often have many joints, so it is im-portant to make the modular design. We should carefully design the modular to increase the reuse rate and robustness. In this subject, we still take Kejia robot as an example.In addition, the service robot also need to make friendly interactions with people, and we discuss that from the perspective of evolutionary psychology.The second part discusses a advanced robotics technique-soft robotics. With the development of robotics, the concept of robot begins to change. In recently years, some robots which are fully-soft and without any joints become to appear. Compared with traditional hard robot, the soft robot often has a better flexibility and safety to the environment. From bionic point of view, this part proposed a new type soft robot called Honeycomb PneuNets (HPN). After some explanation to this concept, we also made a basic analysis of its motion capabilities. The experiments showed that this kind of robot has a good extending rate and bending rate. Further more, we also found that it has a changeable stiffness. Though these analysis and experiments are still preliminary, one can see from that it is a promising technology for soft robotics.The third part discusses a high-performance, low-cost planning technique using smart sensor and Ubiquitous computing technology. Planning algorithms have been developed for a long time, and in recent years a lot of good algorithms become more and more applied. But how to improve the efficiency of them is still a important challenge. In this dissertation, we consider a more specific scene-random objects in random place. We proposed a planning technique using smart cloth, and analyze it to greatly improve it.The main contributions and innovations of this dissertation are listed as follows:Firstly, it introduced a high performance service robot-Kejia2b. This robot is aiming at common tasks of home service robot (such as grasping and moving). The dis-sertation discussed the ideas of the interaction design and DOF placement and analyzed why these design is good for controlling and operating. An innovative elevator system makes it easy to adjust the height of its body. The modular, light weigh manipulator moves very fast and flexible. Kejia2b robot has won the champion of RoboCup and these designs played a big role to the author. So the author believe that these thinking will give new robot designers some inspiration and reference.Secondly, the dissertation proposed a new fully-soft robot. We proposed a proto-type made of metal sheet and air bags. The prototype shows us a very flexible deforma-tion capability. We also discussed the changeable stiffness of HPN and verified it by an experiment. Though these experiments and analysis are still preliminary, one can see from them that the movements of HPN robot is very different from existing soft robot actuators. Compared with previous soft robot actuators, HPN is one of the most flexible actuators and thus may be a new promising technology for soft robotics.Lastly, on the indoor planning concern, this dissertation proposed a new planning technique. Smart cloth is used for object detection and a new algorithm-Exponential Backoff-Sampling Rapid Random Trees (EBRRT) is proposed based on Rapid Random Trees. The main idea in EBRRT is to estimate the most efficacious sampling area to avoid useless sampling. In our experiments, we constructed a map with randomly placed obstacles. In the map, EBRRT and several currently best algorithms were compared through solving a path planning problem. The results showed that our method makes an order of magnitude improvement than the existing ones and the low time consuming characteristic makes it a promising technology for smart devices.