Challenge Description

The main challenge of the Mobile Manipulation Track at the 11th RGMC is to come up with a fully autonomous solution for a multiple pick & place problem that includes tasks that need to be solved with two arms.

Objective¶

There has been an increasing interest in mobile manipulators in many industrial sectors, including manufacturing, agrifood, healthcare, etc. All such use-cases include manipulation of small and light objects and handovers. Mainly tasks that are low-added value but repetitive. This challenge is not intended to solve such problems but to give give a blueprint of a solution by tackling model tasks.

The main objective is to solve a pick & place and manipulation problem that fully exploits the capabilities of a bimanual mobile manipulator.

Tasks¶

The tasks to do can be grouped according to the two main capabilitis of a mobile manipulator

The robot will start somewhere in an arena of 4x6m (tentatively) that it cannot leave. There will be tables in the scenario, one designated for picking, the other for placing the objects. There will be multiple manipulation challenges.

The robot should be able to place itself confortably in a way that it can perform the full corresponding manipulation task without needing to move. There will not be any markers in the scenario hence the so-called docking problem would be place itself solely by means of a visual target.

Manipulation tasks¶

Once in place, the contestants will be challenged with tasks grouped within a Task Board designed by NIST. This Task Board would be designed especially for this very challenge, as done for previous editions of the RGMC. It would include

Accuracy tests by means of laser and a retroreflective target
Picking & placing of objects with non symmetric mass distribution (e.g. a hammer)
Mounting challenges that require both arms of the robot.

Tasks execution¶

Contestants will be challenged to perform all the manipulation tasks in a provided order. Each of the tasks will require to move the robot in the arena and to place it somwhere where the corresponding manipulation task is completely doable. For instance:

For picking an object, once the robot stops, it cannot move further to relocate to grasp more confortably
To do tests that would include placing the arm in various positions on the table (picking & placing, or pointing to different points for accuracy tests) the robot should not relocate.

Success¶

A fully successful run would be to

Perform all the tasks in the sequence of tasks successfully, depending on the definition of these tasks,
Moving towards each of the targets and placing the robot without the need to relocate once stopped
To do the whole sequence within a specific time interval.

Failure¶

Failures (that would lead to points reduction) would be

Failing in performing the tasks, e.g. not being able to pick an object, fallen objects on the floor, not being able to mount what is asked.
Relocating after stopping for a task
Not performing all the tasks in the sequence