Enabling closed-loop control remains a clear challenge for soft manipulators. Going beyond open-loop control basically requires integrating: robust, but soft, sensors to provide multiple modes of proprioceptive and exteroceptive feedback; and, effective control strategies based on deformation modelling or complex learning algorithms.
It is difficult to accurately predict the response of a soft robot due to certain driving condition based on modelling because of the complex behaviours (nonlinearity, hysteresis, viscoelastic effect, large strain, or deformation) of these hyperelastic materials. In the last years, novel sensing solutions have been integrated into different types of actuators, like: omnidirectional actuators to monitor both bending angle and direction; McKibben to measure the contraction length or circumference; bending actuators to measure the bending angle, etc. Despite these attempts, to date, the reconstruction accuracy is still poor (particularly for large deformation). Furthermore, twisting and elongation in soft continuum robots are neglected due to their limited sensing capability. More sensing nodes, better sensor configuration, new modelling and reconstruction algorithms are needed to fully solve this problem. Soft control of robotic manipulators suffers from all these limitations. Moreover, most soft grippers do not include tactile sensing (e.g. granular jamming) but exteroception seems largely necessary for implementing future skilled tasks (e.g. object discrimination), for reacting to unexpected situations (e.g. slippage detection), and appropriately interacting with humans.
From a control perspective, in the last years both, model-free and model-based approaches, used for the control of soft manipulators, have improved the dexterity of these platforms, but demonstrated some limitation in the accuracy or in the interaction with the environment. The possibility to estimate the manipulators’ configuration, relying on an effective sensor implementation, is essential to improve their performances and remains an open challenge.
This workshop aims to provide an insight into the various technologies for soft sensing, sensorized soft actuators, and methods for control of soft robotic manipulators as a guideline for future applications in the soft robotics field.
The final goal will be to bring together researchers from many areas (like control, mechanical and electronic engineering, material science, computer science, etc.) in order to explore how to maximize the progress evaluating the advantages of sensing and their effects on the control strategies.
Topics of interest will include:
- Proprioceptive sensors for soft robots
- Hyperelastic materials for mechanical sensing
- Biomimetic and soft tactile sensing
- Soft actuators with integrated sensing
- Reliable soft actuation strategies for closed-loop control
- Kinematic, dynamic control
- Hysteresis, Friction compensation
- Data driven methods/model free control
- Adaptation in unstructured environments
