Enabling Operational Multilift through Vision-based Control, Scaling, and Adaptation
Abstract
Transport of externally slung loads has been a critical capability of rotorcraft for several decades. Slung load transport is normally done using a single helicopter with one or more cables attached to the payload. However, this limits payload controllability and limits the maximum mass that can be carried to the maximum capability of a single aircraft. Of course, one can develop larger rotorcraft, but it is inevitable that a larger payload (outside the capabilities of available helicopters) will need to be carried. An approach to solve the challenge of “long tail” payloads is to employ multilift: coordinated transport of a slung load by a team of rotorcraft. Multilift has been the subject of research over the past decade. The PIs have previously developed and demonstrated (both indoors in a motion capture studio and outdoors using RTK GPS) a hierarchical load-leading multilift system that: (1) is scalable in both vehicle/payload size and number of vehicles; (2) is able to estimate payload mass properties in flight; (3) can plan payload trajectories that balance cable tensions (and thus control effort) between vehicles in the team. Work done in support of that research includes simulation and analysis to assess controllability as a function of formation and on-line cable tension estimation. The current state-of-the-art in multilift consists of demonstrations using small (kilogram-scale) UAS and payloads and dependence on GPS. Although multilift has been well-demonstrated in laboratory scale research, development work is still required to enable deployment. This includes scaling to operationally useful vehicle and payload masses as well as enabling operations in GPS-denied environments. The research proposed here seeks to address those challenges through: (1) vision-based payload state estimation and navigation; (2) development of scaling laws; (3) enabling operations in complex disturbance environments such as ship airwakes. This will be performed using a combination of analysis, simulation, and hardware demonstrations at lab-scale using small commercially available multirotors. The expected outcome of the research is threefold: first, a vision-based cooperative estimation pipeline to enable multilift in GPS-denied environments; second, controller design specifications (bandwidth and damping) for multilift-capable UAS; third, a machine learning based approach to capture unmodelled aerodynamic disturbances and adaptive control. Collectively, these three outcomes will increase the readiness of multilift for eventual deployment.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- May 08, 2024
- Source ID
- N004212410001
Entities
People
- Junyi Geng
Organizations
- Pennsylvania State University
- United States Navy