Collaborative Tasks Benchmark (CT Benchmark)

Introduction

Benchmarks are crucial to help robotics research advance in an efficient and standardized manner, however, the collaborative assembly field lacks a benchmark with multiple tasks of varying task complexity. As such, we propose the Collaborative Tasks Benchmark (CT Benchmark). It contains 7 distinct tasks ranging from easy to difficult assembly scenarios, each with 1 to 3 parts, providing a rich environment for studying collaborative assembly systems. Its design was grounded on the following criteria:

• Contain independent sequential assembly tasks
• Tasks have a wide variety of sturdy and graspable 3D printable pieces
• Each piece has its own distinguishable features.

The benchmark mainly focuses on assisting in the development of collaborative systems, though it is also applicable towards any combination of its inherent sub-systems: task sequencing allocation, fully-robotic assembly, robot pick-and-place operations and visual perception.

Task description

The proposed benchmark is representative of a city landscape and contains a total of 7 unique tasks. The main assembly operations are wide tolerance insertion (WTI), tight tolerance insertion (TTI), screw fastening, snap fitting and two-handed actions.

| Task | Color | # sub-tasks | # parts | WTI | TTI | Screw | Snap-fit | Two-handed | | :------: | :---------------------------------------------------------------------------------------------: | :---------: | :-----: | :---: | :---: | :---: | :------: | :--------: | | Museum | | 1 | 1 | | | | | | | Snap | | 3 | 3 | | | | | | | Bridge | | 3 | 3 | | | | | | | Triangle | | 1 | 1 | | | | | | | Wheel | | 3 | 2 | | | | | | | Dovetail | | 2 | 2 | | | | | | | Hospital | | 4 | 3 | | | | | |

Benchmark deployment

Essential hardware	Optional hardware
3D printed buildings and base: 520g of PLA 25 hours Fasteners: M5 x 30 screws (x5) ISO 4032 M5 nuts (x5)	3D print corner pieces: 20g of PLA 1 hour Fasteners: M4 x 20 screws (x4) Adequately sized M4 T-nuts (x4)

Additional material

In addition to the essential and optional hardware, three supports are made available to aid the assembly process. These supports were used to guarantee a repeatable setup in the accompanying article while performing the proposed manufacturing assembly scenarios.

Cite our paper

If you've found the CT Benchmark useful for your research, please cite our paper as follows:

@article{duarte2024benchmarking, title = {Benchmarking human-robot collaborative assembly tasks}, author = {Laura Duarte and Miguel Neves and Pedro Neto}, journal = {Results in Engineering}, volume = {22}, year = {2024}, issn = {2590-1230}, doi = {10.1016/j.rineng.2024.102042} }