An Aerial Manipulator for Robot-to-robot Torch Relay Task: System Design and Control Scheme

TL;DR

The paper tackles autonomous robot-to-robot torch relay in ice and snow by designing an aerial manipulator that combines a quadrotor with a 3-DOF manipulator and a vision-guided lighting system. It details hardware choices (multirotor base with serial arm, ski-jump-inspired appearance, LEDs with IR filtering for robust pose estimation, and a cabin heater for low temperatures) and a cascade control scheme with disturbance estimation to manage dynamic coupling between the UAV and the arm. Key contributions include EPnP-based target pose estimation, end-effector control via inverse velocity kinematics, and outdoor validation through torch lighting experiments and a complete relay with an amphibious robot, demonstrating autonomous operation in a real Olympic context. Experimental results show hovering accuracy around $\pm 10$ cm and end-effector tracking around $\pm 2$ cm under outdoor cold conditions, enabling reliable torch lighting and relay capability in the ice and snow field.

Abstract

Torch relay is an important tradition of the Olympics and heralds the start of the Games. Robots applied in the torch relay activity can not only demonstrate the technological capability of humans to the world but also provide a sight of human lives with robots in the future. This article presents an aerial manipulator designed for the robot-to-robot torch relay task of the Beijing 2022 Winter Olympics. This aerial manipulator system is composed of a quadrotor, a 3 DoF (Degree of Freedom) manipulator, and a monocular camera. This article primarily describes the system design and system control scheme of the aerial manipulator. The experimental results demonstrate that it can complete robot-to-robot torch relay task under the guidance of vision in the ice and snow field.

Figures (10)

• Figure 1: Scenario of Olympic torch relay between robots.
• Figure 2: Illustration of competitor's aerial style of ski jumping and aerial robot's body shape: (a) lateral view and top view; (b) tilt top view.
• Figure 3: Aerial manipulator system: (a) system components; (b) structure of vision module; (c) target robot with makers.
• Figure 4: Illustration of the frames, configuration, and task workspace of the aerial manipulator system.
• Figure 5: Software architecture.