Open record · Launch benchmark · Open call

Benchmark #001 Dexterous Manipulation

Final venue and dates will be published after confirmation.

A robotic gripper holding a test object
01 / Purpose

A public starting point across hardware, perception, control, models and task design.

Dexterous manipulation can produce compelling demonstrations, yet object variation, tool variation and repeated operation quickly reveal system boundaries.

It is a useful first test of RobotBay’s public method: the task can be described, conditions made visible, and failure and recovery recorded.

02 / Initial task briefs
Task A

Pick and Place

Test grasping, moving and stable placement across different shapes, materials and starting positions.

Task B

Tool Handling

Test grasping, reorientation and use of a simple tool to complete a defined action.

Task C

Variable-Object Sorting

Test recognition, grasping and classification when the object set is not completely fixed.

These are initial directions. Final task specifications will be published after confirmation with participants and real-world operators.

A robot performing a precision task at a workbench
03 / Evidence dimensions

Show the evidence separately. Do not manufacture an opaque score.

  • Task completion
  • Repeatability
  • Human intervention
  • Setup time
  • Failure mode
  • Recovery behavior
  • Declared constraints
04 / Public record

A record that can be inspected, reviewed and understood in context.

Compare two systems using their full attempt records

  1. 01Task brief
  2. 02Test protocol
  3. 03Test environment
  4. 04Complete attempt record
  5. 05Representative success and failure video
  6. 06Observations
  7. 07Operational boundaries
  8. 08Conflict disclosure
Researchers observing a robotics benchmark test
05 / Initial cohort

The initial cohort is designed for up to five manipulation systems.

Participation is not confirmed until the public record is updated. Unconfirmed brands, names, venues and dates are not displayed.

06 / Growing series

#001 is only the beginning.

Future directions may include mobility, perception, interaction and field operations. Numbers, titles, participants and dates will not be invented before scope and method are confirmed.