This project aims to develop a novel local path planner that ensures the safety and optimality of the movement of autonomous ground vehicles in an unknown dynamic environment.
Autonomous ground vehicles are gaining significant importance in daily life. It can be used as a waiter, assistant for older or disabled people, and delivery of shopping or food from restaurants. It is also common in modern factories and warehouses, where it is used for transportation. These vehicles cooperate with humans or other vehicles. The state-of-the-art local path planning algorithm focuses mainly on collision-free approaches, which is crucial for proper operation. However, nowadays, the energy efficiency and optimality of the process have gained significant attention, i.e., minimizing energy consumption and minimizing task execution time. For this reason, the novel local path planner should be proposed to achieve both: safe operation in a dynamic unknown environment and optimality in terms of smooth movement, shortest path or operation time, and minimizing energy consumption required for task execution.
Task 1. Development of the SEE path planner for an unknown dynamic environment.
Based on the analysis and verification of the algorithms, the most promising ideas will be further developed to achieve the efficient local path planner requirements. As was mentioned earlier, the efficient local path planner should provide the following:
- collision-free operation,
- smooth movement,
- local minima avoidance mechanism,
- goal reaching in narrow passages,
- high energy efficiency,
- short time required to achieve the goal.
Task 2. Simulational verification of the proposed path planner.
Simulation environment (Gazebo software) will be used in the development task to verify the methodologies related to satisfying all the abovementioned requirements.
Task 3. Experimental verification of the proposed path planner.
This task is related to implementing the developed algorithm in a mobile robot for experimental verification. It will also be compared with the solutions from the literature. The realization is partially parallel to the development and simulation verification of the SEE path planner due to the laboratory’s required preparation to provide complex scenarios.
Task 4. Development of automatic selection procedure of SEE path planner parameters.
The parameters of the SEE path planner proposed in Task 2 will be analyzed, and next, the automatic tuning procedure will be proposed. It is assumed that the project investigator will use his experience with nature-inspired optimization algorithms to provide the optimal solution. The proper objective function definition has a significant meaning in the optimization problem. The unbalanced function may give a concept error: the minimum of the objective function may not lead to obtaining a required AGV behavior. Next, the constraints must be determined to provide safe and implementable parameters. The nature-inspired optimization algorithms can solve complex multi-objective optimization problems with constraints relatively quickly.
Task 5 Experimental verification of the proposed automatic tuning procedure.
This task will provide experimental verification of the tuning procedure of path planner parameters in the case of optimality criteria. These criteria will include the following:
- successful goal-reaching,
- goal-reaching time,
- path length,
- consumed energy,
- the smoothness of the movement,
- behavior in a case of dynamic obstacle existence,
- behavior in a matter of simple and complex local minima existence.
Task 6. Preparation of the SEE path planner implementation to deployment to open-source repositories.
Preparing the developed SEE path planner for deployment is a crucial project summary. It will promote the developed path planner to the Robot Operating System society, increasing the impact on the research field and team recognition within this area of research.
In progres...