Robotics-2: Dynamics and movement control

Yulia Sandamirskaya

In this session, we will discuss the role of neuromorphic computing in movement control of articulated robotic systems, such as robot arms. The discussion will start with several short presentations (statements) from INRC researchers and quests who work in domain of arm or full-body control. These presentations will include work in arm control that has been done on Loihi so far, an overview of mechanisms behind human arm control, and examples of efficient bio-inspired control of a bio-inspired robotic body. Thus, our discussion will unfold between the neuromorphic versions of conventional control (PID, MPC) and bio-inspired control strategies, for which a neuromorphic implementation is yet to be established. Along the discussion, we will try to find answers to the following questions:

• Which subtasks or levels in motor control are particularly attractive for neuromorphic solutions and which are not? What would be the criteria for success of a solution? Which bottleneck could neuromorphic realization solve: scalability? latency? throughput?

• What are the challenges to make Neuromorphic approaches enter the main stream in robot control: Algorithms? Interfaces? Performance? Ecosystem?

Other questions can be posted in our slack channel #robotics.

Since the keynote presentations will be ultra-shot 5-7 minute statements, please, have a look at the related previous talks of the speakers, listed below.

• Elishai Ezra Tsur (Neuro-Biomorphic Engineering Lab, Open Univ. Israel, http://nbel-lab.com/ )

  • Neuromorphic NEF-based inverse kinematics and PID control

  • INRC Forum talk: Adaptive neuromorphic control for assistive robotics

• Florian Walter (Robotics, Artificial Intelligence and Real-time Systems, TUM, https://www.in.tum.de/en/i06/home/ )

  • Neuromorphic Computing is a Key Enabler for Biomimetic Robotics

  • INRC Forum talk: Brain-Derived Adaptive Control of Biomimetic Robots with Intel Loihi

• Manos Angelidis (Neuromorphic Computing, fortiss, https://www.fortiss.org/en/research/fields-of-research/detail/neuromorphic-computing )

  • Nonlinear swimming robot control with Loihi

  • INRC Forum talks: A spiking Central Pattern Generator for the control of a simulated lamprey robot ; Nonlinear swimming robot control in Loihi

• Gregor Schöner (Institute for Neuronal Computation, Ruhr-Univ. Bochum, https://www.ini.rub.de/ )

  • Key lessons from human movement  for neuromorphic approaches to movement generation

• Philipp Stratmann (Robotics and Mechatronics, German Aerospace Center, https://www.dlr.de/rm/en/desktopdefault.aspx/tabid-8017/ )

  • Learning to harness elasticity for optimal locomotion

• Alona Kharchenko, Rafael Hostettler (Roboy, Munich, https://roboy.org/ )

  • Roboy 3.0: bio-inspired musculoskeletal humanoid robot platform. Challenges in classical control and opportunities for neuromophic approaches

A collection of robotics-related talks from last year’s INRC Forum can be browsed here: Tutorials and Related Presentations

Not yet available

Not yet available