In this paper we propse a construction method of autonomous tomato harvesting robot. For tomato harvesting we focus to harvesting device and visual recognition method. We propose 2 types of harvesting method; cutting scissor and rotational plucking gripper. For recognition, we compare different 3 types of 3D camera, then we construct tomato fruit recognition method using color detection and sphere shape fitting, and also propose tomato pedicel detection method by robust estimation using direction from center of fruit to pedicel and gradient orientation of pedicel edge. Combining these elements, we construct some autonomous harvesting systems and evaluate harvesting of real tomatoes in the tomato robot competition and the real farming veniel house. According to evaluation results, both cutting and plucking has different advantage and disadvantage, it is important to select appropriate one for breed or individual. Also for recognition, pedicel recognition, which must be broken, increases reliability harvesting.

In this paper, we described a developed life-size humanoid robot platform. A purpose of the development is to build a robot platform which can continue to work even when it falls. To realize continuous operation while contacting with the environment, we designed to use linear actuators in order to endure impact force to joints. Tough contact points were located to support self-weight while contacting with environment. we present a methodology to create a motion which a robot contacts to environment without failure by using the tough contact points. Proposed method was confirmed by the experiment that the robot fell down without mechanical failures and continued to work after falling down by using tough contact point.