Design and Implementation of a Haptic Backhoe Testbed

Abstract submitted to 2004 ASME International Mechanical Engineering Congress

The advent of a new generation of haptic input devices has opened up new possibilities for the fluid power industry. A novice can now become proficient at operating heavy earthmoving equipment much more quickly than was previously possible with the use of haptic control interfaces.
 

The operation of heavy equipment such as loaders and backhoes has previously been reserved solely for professional operators with the required training and experience. It takes time to acquire a “feel” for the non-intuitive lever motions necessary to load a truck or dig a trench effectively and efficiently, because the command motions actuate joint space variables and both the desired end-effector trajectory and visual feedback exist primarily in Cartesian space. Even more difficult is the ability to sense the forces experienced by the end-effector when the only feedback available to the operator consists of the observed bucket speed through the soil, the engine’s response to a load, or pressure waves propagated back to the user’s hand through the valve spool and control lever.
 

Therefore, a novice operator with an earthmoving task has heretofore been forced to either hire a professional, which can be expensive especially for a small job, or rent the equipment and learn how to use it him or herself before the task can be completed. However, if the traditional direct-acting valve control levers are replaced with a haptic interface, and appropriate valves and joint position sensors are installed, three significant improvements result. First, kinematic transformations can be performed as part of the real-time control loop so that the operator thinks and works solely in Cartesian space. Second, representations of forces experienced by the end-effector can be displayed to the users hand via the active nature of the haptic interface. Third, controls can be physically separated from the equipment and teleoperated if desired because the mechanical lever commands have been replaced with electric signals.
 

To prove this concept, a haptically operated backhoe has been developed at the Georgia Institute of Technology for testing and evaluating control algorithms, where the master haptic input device is the PHANToM 1.0 available from Sensable Technologies, the slave is a John Deere model 47 backhoe, and control is implemented with host and target computers running Matlab/Simulink/xPC Target software available from Mathworks. Electrohydraulic valves from Sauer-Danfoss and magnetorestrictive position sensors from Balluff have also been retrofitted to the backhoe.
 

In the controller, real-time computations are required for several quantities, including the kinematic transformations between three vector spaces, (cylinder, joint, and Cartesian), master/slave position errors, and tip force representations. This paper will present one such control scheme, providing a suitable algorithm for real-time operation of the system dubbed the Robotic Backhoe with Haptic Display.