CA-INP contributes to different WP in the project.
For WP4, we designed a dual quaternion based dynamic movement primitives (DMP) to learn bimanual deforamation tasks.
We compared it’s performance with traditional decoupled DMP. We also added shape feedback to improve shape servoing. These methods are integerated in CA-INP mockups.
For WP5, we developed further our tracking-servoing approach to be applicable in different cases. We also integrated our approach with the SoftManBot #software to be tested with Michelin case. Furthermore, for doll use-case, a 3d reconstruction effect is implemented using the signal from DIGIT tactile sensor.
Based on this 3d tactile feature, more manipulation experiments with doll pieces are being performed.
The #work about grasping is presented in ICRA2022. Furthermore, CA-INP was invited as a speaker in ERF2022 conference to present SoftManBot project.