Active Robot Perception for Automated Potato Planting

该项目研究了新的主动机器人感知解决方案，用于泽西岛农民的马铃薯种植自动化。将探索从存储的马铃薯种子的低成本传感和质量控制技术，包括机器人机械手上的摄像头，以识别和估计需要由抓取末端执行器处理的马铃薯的姿态。将在机器人传感系统中设计和实施适当的软件解决方案，以合理准确并适用于真实的户外场景。该项目还将探索结合联合收割机视觉和触觉信息（即摄像头+触摸传感器）的可能性，以改善对板条箱中马铃薯的检测和处理。机器人感知系统将集成到最终的移动的平台中。

The project investigates new active robot perception solutions for the automation of potato planting at Jersey Farmers. Technologies for low-cost sensing and quality control of seed potatoes from the storage will be explored, including a camera on a robot manipulator to identify and estimate the pose of potatoes that need to be handled by the grasping end-effector. Opportune software solutions will be designed and implemented in the robot sensing system to be reasonably accurate and suitable for real outdoor scenarios. The project will also explore the possibility to combine visual and tactile information (i.e. camera + touch sensor) to improve the detection and handling of potatoes in crates. The robot perception system will be integrated to the final mobile platform.