Background
Impact-aware robotics is a very timely topic, the relevance of which is also testified by the dedicated Special Issue recently opened by IEEE T-RO, one of the most important scientific journals in robotics: https://www.ieee-ras.org/publications/t-ro/special-issues/impact-aware-robotics.
The reference trajectory of a robotic system subject to impacts contains jumps in the reference velocity. Tracking such trajectory in the real system is very challenging because the closed-loop system will experience the impacts at different times than the one corresponding to the pre-planned reference velocity jumps, due to the unavoidable uncertainty of the real world. This results in huge control errors in the time instants close to the expected and actual impact. A way to avoid this drawback is called reference spreading [1]. The method has been applied to 1-DoF systems and extended to robotic arms [2] and humanoids [3]. However, its applicability to aerial robots has never been studied to date.
Assignment Description
This assignment is aimed at exploring a new field, that we may define as impact-aware aerial robotics. The work will first focus on analysis, through a deep understanding of the related state of the art, of the following points:
- Motivations behind impact-aware aerial robotics
- Specific challenges of aerial robotics compared to the other robotic fields considered so far
- Selection of interesting use-cases
Second, the assignment will select a simple enough model of an aerial robot through simulations in Matlab/Simulink, Ros/Gazebo, or other setups (the availability of impact modeling should be taken into consideration when selecting the simulation tools, and the state of the art can also shed light on this practical point). The reference spreading method will then be implemented on the aerial robot model for validation. If the previous points are satisfactorily completed ahead of time, hardware experiments could be also considered.
The assignment research questions are formulated as follows:
- Is an impact-aware approach interestingly motivated by an application in aerial robotics?
- What distinguishes the aerial robots in this scenario from other robotic systems, i.e., which are the area-specific challenges that arise when applying impact-aware control methods to aerial robots?
- Can reference spreading be successfully applied to an aerial robot subject to impacts?
References
[1] Mark Rijnen , Alessandro Saccon , and Henk Nijmeijer “Reference Spreading: Tracking Performance for Impact Trajectories of a 1DoF Setup” IEEE Transactions On Control Systems Technology, Vol. 28, No. 3, May 2020
[2] van Steen, Jari J., Nathan van de Wouw, and Alessandro Saccon. "Robot control for simultaneous impact tasks via QP based reference spreading." arXiv preprint arXiv:2111.05211 (2021).
[3] Rijnen, Mark, et al. "Control of humanoid robot motions with impacts: Numerical experiments with reference spreading control." 2017 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2017.