ZürichAutonomous Systems Lab

Magnetic wheel Main problem: Solution: active rotaryconcave edges lifter mechanism

Obstacle passing principle

Wheel configuration• Motorbike wheel arrangement with free joint in the fork surface adaptation• Lifter arms used as lateral stabilizers when driving sideways

Magnebike: Compact Magnetic Wheeled Robot

Locomotion Concept

• Explore and navigate in industrial plants with mobile robots.

• Improve the existing inspection technologies by building compact robots with high mobility.

• These robots will allow bringing inspection sensors (camera, ultrasonic probes, Eddy current probes … ) to locations hardly or not accessible by humans.

• Gaining inspection time and increasing the inspection performance using robots: e.g. no need to disassemble the part to be inspected or to move it to a workshop.

Fabien Tâche, Wolfgang Fischer, Gilles Caprari, Roland Siegwart, contact: ftache@ethz.ch http://www.asl.ethz.ch/research/asl/alstom

SpecificationsMotivation• Wide range of inner diameters: from 200mm up to 800mm.

• Local abrupt inner diameter changes: 90° convex and concave obstacles.

• Complex arrangement and sequence of obstacles such as triple step or gap.

• Any inclination can be encountered: climbing ability is required.

• Maneuver about in narrow locations, travel on circumferential paths.

• Material: ferromagnetic.

Active Control: wheel speed control• Measure robot deformation assign front/rear wheel speeds

Active Control: lateral stabilization mechanism• Measure contact force assign lifter arm position

Results: tests in real environment

The user does not see the robot and the environment model is unknown. The goal is to provide visualization tools that help the user to remote control the robot.

Strategy: SLAM (Simultaneous Localization and Mapping)

… move (odometry) 3D scan scan matching move (odo) 3D scan ..

• 3D scan: provides a local 3D model of the environment.

• Odometry: used to track the robot position between 2 consecutive 3D scans. Used as prior knowledge for the scan matching.

• Scan matching: determines the transformation matrix (3D robot movement) between 2 consecutive scans, but also generates a global 3D model of the environment.

Localization Concept

Sensor: small size lightweight 3D range finder

Sensor: rotating Hokuyo-URG 3D scan in real environment

3D scan matching: using ICP (Iterative Closest Point) algorithm

3D CAD model of a typical environment

200mm800mm

g

Triple step

Gap

Fmag

Fmag

Fmag

Fmag

No control With control

No contact Stabilizing Lifting

Robot characteristics• Mass: 3.3kg• Size: 170x130x220mm3

• Wheel diameter: 60mm• Mag. wheel force: 250N• Max. speed: 2.7m/min• Operating voltage: 24V

Objectives• Locomotion: develop new climbing robots with high mobility able to access any

location in narrow environments with complex geometry.

• Localization: develop a robot as autonomous as possible, in order to simplify its control by the end-user: develop high level navigation strategies.

• Integration: integration of locomotion mechanism, localization sensors and control electronic in a compact system

Acknowledgments