Sub-Project 4: Development of Running and Jumping Humanoid Robots Based on Bio-Inspired Algorithms and Deep Learning

The main purpose of the integrated project is to design and implement running and jumping and/or flying robots, which are humanoid and grasshopper-like robots. The mechanism will be designed by imitating animals which have outstanding performance in running and jumping and/or flying by adopting algorithms to reach the optimization. We will adopt various sensors and learning algorithms in order to increase the learning ability of robots. As a result, robots can not only complete tasks by combining running and jumping but also modify moving trajectory in the same time to reach the optimized movement, and robots will have wider applications. The tasks of each year are list as follows.First year: We will focus on developing the mechanism which is based on animals that have outstanding performance on running and jumping and/or flying motions. We will also analyze running and jumping in details to build control system. Second year: The learning algorithms and optimization methods will be developed, and the designed robot will be able to jump, run and/or fly. Robots can learn jumping and running from artificial intelligence algorithms. The performance can be also validated in various robot competitions.Third year: The research focuses on combining sensors and artificial intelligence algorithms on practical applications, such as integration of robot pose and path planning, and movement decision making based on artificial intelligence algorithms to make sure that the robots can perform tasks independently. Abstract of Sub-Project 4: This research will first focus on the analysis of the relationship between the structure of legs and body and movement such as running and jumping. Secondly, the concept of conservation of angular momentum will be applied to keep robots balance. Lastly, various sensors will be installed on the robot to measure center of gravity and angle of inclination while jumping and running and combine with path planning system to decide its movements. In the end, the biped jumping and running robot which has better agility and weight-bearing ability will be investigated. The tasks of each year are list below:First year: The designed structure can reduce the impact force will be designed by the result of the above analysis. Also, we will adjust ratio of the robot to increase the efficiency of movements.Second year: In order to increase the applications of jumping and running robots, this research will focus on the development of upper body, such as arm and weight-bearing mechanism design. Various sensors such as depth camera, inertial measurement unit (IMU) and pressure sensor will be adopted so that the robot will have the ability to acquire the environmental information.Third year: The research will focus on the optimization of robot’s poses, and then the robot can acquire its pose by installing pressure sensors. After the robot gets the information of environmental information and its poses, we will combine bio-inspired algorithms and a series of DDPGs to design movement and path planning system. As a result, this jumping and running robots will have the ability to accomplish tasks independently.