Vision-Based Grasping Benchmark

Benchmark results

We display the performance of different grasp-generation methods on the VisGraB benchmark. Since different grasping methods may have their own strengths and weaknesses, we do not summarize the results in a single value. Instead, we give the distribution of grasping results for the different conditions. Furthermore, the grasping quality for each condition is given as the average lift quality and the average grasp wrench-space quality over the successful grasps.

GroupShort descriptionResults
Popovic et al. Three grasp-generation methods based on surface information extracted from the Early Vision System. results

Gert Kootstra, Mila Popović, Jimmy Alison Jørgensen, Danica Kragic, Norbert Krüger