Reliable motion planning for nonlinear systems under uncertainties (REPLAN) is a “Young Team” research project funded through UEFISCDI with these details:

• registration code PN-III-P1-1.1-TE-2019-1614
• contract number TE27/2020
• 24 month duration (September 2020 – August 2022)
• 431 900 RON funding (approx. 88 000 euro)

Short description

The REPLAN project proposes various procedures for the reliable computation of trajectories in a multi-agent framework (i.e., offline trajectory generation, followed by trajectory tracking and reconfiguration) in the presence of realistic disturbances and model uncertainties.

The main idea is to exploit theoretical tools (flatness, NURBS functions, set-theoretic elements) in order to obtain a collection of trajectories which respect the internal dynamics and the operational constraints (collision avoidance, coverage restrictions, target tracking, etc.) of the agents while in the same time minimizing the given costs (e.g., time to destination, energy consumption, path length, etc.).

Objectives

The main objectives of REPLAN are:

1. Offline trajectory generation via flatness with NURBS parametrizations (WP1)
   • Analytic descriptions of constraints and costs
   • Order reduction for complex representations
   • Heuristic methods
2. Improvements in reliability and stability guarantees under model variation (WP2)
   • Robust design for model variation
   • Adaptive design for model variation
   • Time variation along the trajectory
3. Validation in complex control scenarios (WP3)
   • Simulations without control reconfiguration
   • Simulations with control reconfiguration
   • Validation on experimental setups

Deliverables

• Project website on which relevant information will be updated periodically.
• Technical report gathering algorithms for trajectory generation (trajectory generation for the nominal case with bounded disturbances, model reduction, computation time reduction procedures).
• Technical report gathering algorithms for trajectory generation (trajectory generation for uncertain models; robust and adaptive approaches).
• Control implementation for trajectory tracking in the single/multi-agent agent case.
• Report gathering simulation results (scenarios, implementation code, Gazebo illustrations).
• Report gathering experimental results (scenarios, implementation code, illustrations).
• Toolbox gathering all routines and benchmarks used throughout the project.
• Articles in top journals and conference proceedings (1 journal paper and 2 conference papers per research WP).

Reports

Each phase of the project concludes with the budget and a synthetic scientific report (both in Romanian).

Phase 1 (Sep. 2020 – Dec. 2020)

Phase 2 (Jan. 2021 – Dec. 2021)

Phase 3 (Jan. 2022 – Aug. 2022)