Among the path planning methods, there is a point-to-point path planning, such as dijstra, or algorithms such as A*, which focus on the shortest point-to-point path and tend to be an optimal choice. The other is full coverage path planning. This type of path planning focuses on traversing the entire map, such as sweeping robots. Their main purpose is to traverse. In response to this demand, full coverage path planning has emerged.
Some directions that are often paid attention to in the current full-coverage path planning:
1. Obstacles in the fully covered working area will affect the working path of agricultural machinery, thereby affecting the efficiency of on-site work
2. In the full-coverage operation, there are problems such as omission of the area of the work area and repeated work paths, which in turn affect the quality of work and cause waste of resources. The path planning of the robot can avoid path repetition, shorten the moving distance, and improve work efficiency
3. When faced with complex tasks, the work efficiency of a single robot is low, and many tasks cannot be completed. In order to deal with complex problems and improve work efficiency, multi-robot cooperation is an effective way. The core problem of multi-robot cooperative operations is to solve robot task allocation problem among them.
Therefore, in response to these problems, the current research directions mainly include three aspects: environment model construction, robot path planning, and multi-robot collaborative task assignment.
Accurate and reliable environmental maps are conducive to follow-up path planning. In the case of known environmental maps, static obstacles can be avoided and operational reliability can be improved; robot path planning can effectively reduce the number of turns at the headland, reduce the missed area, and improve operational efficiency; The optimized task allocation scheme is conducive to dealing with complex full-coverage operation problems, and can perform tasks faster and more effectively, reduce operation time, and improve work efficiency.
1. Environment modeling method
At present, in the field of full-coverage path planning research, environmental modeling methods mainly include topological graph method, grid method, geometric graph method and cell decomposition method, etc.
Mainly introduce the cell decomposition method and the grid map method (these two methods are my main focus at present).
Unit decomposition method:
The unit decomposition method (area decomposition method) is to divide the entire space area according to different obstacle shapes, and divide the entire area to be covered into several sub-areas with relatively simple shapes without obstacles and without overlapping parts. Commonly used element decomposition methods include trapezoidal decomposition method, Boustrophedon element decomposition method, and Morse decomposition method.
Common approaches to this type of problem are:
Step 1: Decompose the entire map into units to obtain a piece of unit map
Step 2: Make coverage path planning for each unit map
Step 3: Combine all units, considering the starting point and end point of each unit, combinatorial optimization problem
Grid method:
The grid map method is a method for binary information to represent the environment. The environmental information of the robot workspace is
divided into grids of the same size, and the environmental information of the area is expressed in units of grid units, and then the path is used on the environmental map. The planning algorithm performs path planning. At present, many path planning algorithms use the grid map modeling method.
For the grid method, the obtained grid map will have corresponding values, so that the free area and the obstacle area in the map can be distinguished, and the path planning algorithm can be designed by using this information.
2 path planning algorithm:
The key issue in full-coverage path planning is how to traverse the part of the working area except for obstacles to the greatest extent and effectively avoid
all obstacles.
Simple path planning algorithms include: reciprocating covering method and spiral covering method
Reciprocating coverage method:
This type of method, such as the cattle farming method, is to traverse the area according to certain rules. This type of algorithm is simple, but the omission rate is higher
Reciprocating coverage is easy to implement and has the advantage of high job coverage, but the number of turns is large and 180° turns are required, making turns difficult, resulting in the problem of large turning areas, and it is difficult to implement for larger robots.
Spiral covering method:
This method is spiral like the name, traversing from the outside to the inside or from the inside to the outside.
The spiral type is divided into inner spiral and outer spiral. The inner spiral coverage method is shown in Figure 2b, which means that the robot starts to work from one side of the work area until the area boundary turns 90° to the other boundary to continue.
Optimized Path Planning Algorithm
Biologically Inspired Neuron-Based Approaches
Biologically Inspired Neural Network (BINN) Algorithm
Aiming at the problem of avoiding obstacles and escaping from the dead zone in the process of robot coverage, RASHID et al. correspond one-to-one between the two-dimensional grid map units that require full coverage and the neurons of the biologically inspired neural network, and use the activity value of neurons and the robot’s The last location information is used for real-time path planning. This algorithm does not have a neural network learning process, and the algorithm has good real-time performance.
These heuristic methods or optimization methods can greatly help the path selection.
Hybrid path planning method:
The hybrid path planning method is a combination of simple path planning method and complex path planning method. Simple path planning method or improved algorithm is used to traverse sub-regions, and then combined with optimization algorithms (such as ant colony algorithm, A∗ algorithm, depth-first search, etc.) Carry out connection optimization of sub-regions to achieve full-coverage path planning.
Common optimization methods include theta* method, which can make up for the overly rigid selection direction of the A* algorithm. There are more directions to choose from.
3 Multi-robot task allocation method
Multi-robot task assignment refers to setting an evaluation index for multi-robots, assigning a series of tasks to specific robots, and maximizing the benefits of the robots while the tasks can be fully completed. In the full-coverage operation, the task of each robot is an independent work area divided in the physical environment, and all the robots complete the assigned tasks, which realizes the full coverage of the work area.
3.1 Task assignment method based on mathematical programming
3.2 Task assignment method based on heuristic search
3.3 Task assignment method based on swarm intelligence
3.4 Task allocation method based on market mechanism
paper:
Wang Ning, Han Yuxiao, Wang Yaxuan, Wang Tianhai, Zhang Man, Li Han, Research Progress on Full Coverage Operation Planning of Agricultural Robot