Skip to content

houhuawei23/swarm-uav-task-assignment

BranchesTags

Repository files navigation

Repotitory for Swarm UAV Task Assignment

TODO

  • implement the algorithm in the csci2024@薛舒心, ./solvers/csci.py
    • add time constraints
    • uavs and tasks cluster
  • fix bugs in CoalitionFormationGame task assignment algorithm.
  • Calculate various evaluation indicators, such as:
    • task completion rate, and resource use rate
  • Implement algorithm in iros2024@LiwangZhang, ./solvers/iros.py
  • Implement algorithm in icra2024@LiwangZhang, ./solvers/icra.py
  • Warning: iros2024@LiwangZhang and icra2024@LiwangZhang both need random sample in trigger uav stage! Otherwise, the alg may be vibrate and stuck in a deadlock.
  • Implement Read/Write for solvers test results, using pandas.
  • Test solver on one profile multiple times and draw the box fig.
  • Implement Real Muti-Process simulation in ./solvers/icra.py.
  • Mutiple MRAT Solver Algorithm Implment:
    • Auction Based Methods: Consensus-Based Bundle Algorithm (CBBA) and Contract Net Protocol (CNP)
    • Optimization Based Methods:
      • Deterministic Optimization: linear programming (LP), mixed-integer linear programming (MILP), and the Hungarian algorithm. milp_solver.py and nlp_solver.py
      • Meta-heuristics: genetic algorithms (GA), simulated annealing (SA), and swarm intelligence
    • RL Based Methods
    • Hybrid Methods
  • 算法复杂度分析。。。
  • Implement Overlapping CFG alg.

How to run

该系统用于模拟和测试无人机集群任务分配算法，支持多种测试场景和可视化功能。

系统提供两个主要命令：test 和 plot。

1. 测试命令 (test)

用于运行算法测试，支持多种测试场景。

基本用法：

python main.py test [参数]

主要参数说明：

  • --test_case: 测试用例类型或文件路径
    • 可选值：uav_num、task_num、hyper_params.xxx 或具体的测试文件路径
  • --uav_nums: UAV 数量列表，默认为 [40]
  • --task_nums: 任务数量列表，默认为 [20]
  • --hp_values: 超参数值列表
  • --choices: 要测试的算法列表
  • --timeout: 每个算法的超时时间（秒），默认为 10
  • --random_test_times: 随机测试次数，默认为 5
  • -o/--output: 输出文件路径
  • --show: 是否显示图表
  • --sim: 是否时间步模拟运行（根据任务分配结果，无人机前往任务点并执行任务）
  • --save_dir: 保存分配结果示意图的目录

示例：

# test
python ./main.py test --test_case ./test_cases/case4.json \
--choices csci centralized distributed \
--save_dir ./.image

# 测试不同 UAV 数量下的算法性能
python main.py test --test_case uav_num --uav_nums 20 40 60 --task_nums 20 --choices algorithm1 algorithm2

# 测试不同任务数量下的算法性能
python main.py test --test_case task_num --task_nums 10 20 30 --uav_nums 40 --choices algorithm1 algorithm2

# 测试超参数影响
python main.py test --test_case hyper_params.alpha --hp_values 0.1 0.2 0.3 --choices algorithm1

# 运行特定测试用例
python main.py test --test_case test_case.json --choices algorithm1 algorithm2

python main.py test --test_case ./test_cases/case5.json \
--choices csci iros icra centralized distributed \
--save_dir ./.image

目前支持的算法包括：

choices_to_class = {
    "enum": enum_solver.EnumerationSolver,
    "brute_force": brute_force_solver.BruteForceSearchSolver,
    "csci": csci2024.ChinaScience2024_CoalitionFormationGame,
    "iros": iros2024.IROS2024_CoalitionFormationGame,
    "iros2": iros2024.IROS2024_CoalitionFormationGame_2,
    "icra": icra2024.ICRA2024_CoalitionFormationGame,
    "auction": auction_solver.AuctionBiddingSolverAdvanced,
    "auction_kimi": auction_solver.AuctionBiddingSolverKimi,
    "milp": milp_solver.MILPSolver,
    "milp_pyomo": milp_solver.MILPSolverPyomo,
    "nlp_scipy": nlp_solver.NLPSolverScipy,
    "nlp_pyomo": nlp_solver.NLPSolverPyomo,
    "centralized": centralized_solver.CentralizedSolver,
    "centralized_random_init": centralized_solver.CentralizedSolver_RandomInit,
    "centralized_selfish": centralized_solver.CentralizedSolver_Selfish,
    "centralized_pareto": centralized_solver.CentralizedSolver_Pareto,
    "distributed": distributed_solver.DistributedSolver,
    "distributed_random_init": distributed_solver.DistributedSolver_RandomInit,
    "distributed_selfish": distributed_solver.DistributedSolver_Selfish,
    "distributed_pareto": distributed_solver.DistributedSolver_Pareto,
}

2. 绘图命令 (plot)

用于可视化测试结果。

基本用法：

python main.py plot [参数]

主要参数说明：

  • -f/--file_path: 结果文件路径
  • -x/--xlabel: X 轴标签，默认为 "uav_num"
  • --labels: 要显示的指标列表，默认为 ["elapsed_time"]
    • 可选值：elapsed_time（运行时间）、completion_rate（完成率）、resource_use_rate（资源使用率）、total_distance（总距离）、total_energy（总能量）、total_exploss（总损失）
  • --choices: 要显示的算法列表
  • --save_dir: 图表保存目录
  • --show: 是否显示图表

示例：

# 显示运行时间对比
python main.py plot -f results.yaml --labels elapsed_time --choices algorithm1 algorithm2

# 显示多个指标
python main.py plot -f results.yaml --labels elapsed_time completion_rate --choices algorithm1 algorithm2

# 显示所有指标
python main.py plot -f results.yaml --labels all --choices algorithm1 algorithm2

其他说明

测试结果将保存在 ./.results/ 目录下，文件名格式为 results_{test_case_name}_{algorithm_names}.yaml。

  1. 运行测试前请确保已安装所有必要的依赖包
  2. 使用 --show 参数可以在测试或绘图时直接显示结果图表
  3. 对于超参数测试，必须提供 hp_values 参数
  4. 绘图时可以使用 all 作为 labels 参数来显示所有可用指标

测试结果示意图

Case4: ./test_cases/case4.json

left: Centralized; right: Distributed

left: IROS2024_LiwangZhang; right: ICRA2024_LiwangZhang

CSCI2024_Xue_case4

Ramdom Test and Compare

改变无人机数量：

改变任务数量：

改变超参数: 资源贡献权重 resource_contribution_weight

改变超参数: 路径成本权重 path_cost_weight

消融实验: 最大加权匹配初始化 vs 随机初始化

偏好关系对比实验

Project Structure

1. framework 目录

核心框架实现，包含基础组件和功能模块。

framework/
├── __init__.py           # 包初始化文件
├── base.py               # 基础类和接口定义
├── task.py               # 任务相关定义
├── uav.py                # UAV（无人机）相关定义
├── coalition_manager.py  # 联盟管理模块
├── mrta_solver.py        # 多机器人任务分配求解器
├── sim.py                # 时间步模拟
├── test.py               # 测试框架
└── utils.py              # 工具函数
  • base.py: 定义系统的基础类和接口
  • coalition_manager.py: 负责管理无人机联盟的形成和协调
  • mrta_solver.py: 多机器人任务分配问题的求解器接口
  • sim.py: 系统仿真模块，用于模拟无人机集群的运行
  • task.py: 定义任务相关的数据结构和操作
  • test.py: 测试框架，用于评估算法性能
  • uav.py: 定义无人机相关的属性和行为
  • utils.py: 提供各种工具函数

2. solvers 目录

包含各种任务分配算法的实现。

solvers/
├── __init__.py           # 包初始化文件
├── driver.py             # 求解器驱动模块
├── csci2024.py           # CSCI 2024 算法实现
├── iros2024.py           # IROS 2024 算法实现
├── icra2024.py           # ICRA 2024 算法实现
├── centralized_solver.py # 集中式求解器
├── distributed_solver.py # 分布式求解器
├── auction_solver.py     # 拍卖算法求解器
├── enum_solver.py        # 枚举求解器
├── milp_solver.py        # 混合整数线性规划求解器
├── nlp_solver.py         # 非线性规划求解器
├── brute_force_solver.py # 暴力求解器
└── utils.py              # 工具函数
  • auction_solver.py: 实现基于拍卖机制的任务分配算法
  • brute_force_solver.py: 实现暴力搜索算法
  • centralized_solver.py: 实现集中式任务分配算法
  • distributed_solver.py: 实现分布式任务分配算法
  • driver.py: 负责管理和调用不同的求解器
  • milp_solver.py: 实现混合整数线性规划算法
  • nlp_solver.py: 实现非线性规划算法
  • csci2024.py/icra2024.py/iros2024.py: 特定会议论文的算法实现

  1. 开发新算法时，建议在 solvers 目录下创建新的求解器文件

  2. 修改核心功能时，主要关注 framework 目录下的相关模块

  3. 测试新算法时，使用 framework/test.py 提供的测试框架

  4. 工具函数可以根据需要放在对应目录的 utils.py 中

  5. 添加新的求解器时，需要确保在 solvers/__init__.py 和 solvers/driver.py 中正确注册

  6. 修改核心接口时，需要确保向后兼容性

  7. 建议遵循项目的代码风格和命名规范

  8. 在添加新功能时，注意更新相应的测试用例

  • src/:

    • base.py: base class for the project.
    • uav.py: the class for UAV and UAVManager.
    • task.py: the class for Task and TaskManager.
    • coalition.py: the class for CoalitionSet.
    • utils.py: the utility functions.
    • task_assign.py: implement EnumerationAlgorithm.
    • game.py: implement CoalitionFormationGame.
    • sim.py: the main simulation script.
    • gen.py: generate test data, in json format.

  • tests/

    • case0.json: 3 uavs, 2 tasks; no coalition.
    • case1.json: 5 uavs, 2 tasks.
    • case2.json: 5 uavs, 3 tasks.
    • case3.json: 10 uavs, 5 tasks.
    • case4.json: 50 uavs, 5 tasks.

Reference Commands

# vary uav nums
$ python ./main.py --test_case uav_num --choices all --random_test_times 25 --uav_nums 10 20 50 80 100

# vary task nums
$ python ./main.py --test_case task_num --choices iros csci --random_test_times 25 --task_nums 10 20 50 80 100

# vary hyper params
$ python ./main.py --test_case hyper_params.path_cost_weight --choices csci --random_test_times 20 --hp_values 1.0 2.0 4.0 8.0 16.0 50.0

# test case4.json
python ./main.py test --test_case ./test_cases/case4.json --choices csci

# test uav num
python ./main.py test --test_case uav_num --choices all --random_test_times 5 --uav_nums 10 20 40 80 --task_nums 20 -o ./.results/results_uav_num_all_0329.yaml
python ./main.py plot -f ./.results/results_uav_num_all_0329.yaml -x uav_num --labels all

# test task num
python ./main.py test --test_case task_num --choices all --random_test_times 5 --task_nums 10 20 40 80 --uav_nums 20 -o ./.results/results_task_num_all_0329.yaml
python ./main.py plot -f ./.results/results_task_num_all_0329.yaml -x task_num --labels all

# test hyper params
## 资源贡献权重
python ./main.py test --test_case hyper_params.resource_contribution_weight --choices all --random_test_times 20 --hp_values -10.0 -1.0 0.0 1.0 4.0 8.0 16.0 -o ./.results/results_hyper_params.resource_contribution_weight_all_0329.yaml
python ./main.py plot -f ./.results/results_hyper_params.resource_contribution_weight_all_0329.yaml -x hyper_params.resource_contribution_weight --labels all --show --save_dir ./.results/results_hyper_params.resource_contribution_weight_all_0329

## 路径成本权重
python ./main.py test --test_case hyper_params.path_cost_weight --choices all --random_test_times 20 --hp_values -10.0 -1.0 0.0 1.0 2.0 4.0 8.0 16.0 -o ./.results/results_hyper_params.path_cost_weight_all_0329.yaml
python ./main.py plot -f ./.results/results_hyper_params.path_cost_weight_all_0329.yaml -x hyper_params.path_cost_weight --labels all

## 威胁权重
python ./main.py test --test_case hyper_params.threat_loss_weight --choices all --random_test_times 20 --hp_values -10.0 -1.0 0.0 1.0 4.0 8.0 16.0 -o ./.results/results_hyper_params_threat_loss_weight_all_0329.yaml
python ./main.py plot -f ./.results/results_hyper_params_threat_loss_weight_all_0329.yaml -x hyper_params.threat_loss_weight --labels all

python ./main.py test --test_case uav_num --choices all --random_test_times 25 --uav_nums 10 20 50 80 100 --task_nums 10

## 最大迭代次数
python ./main.py test --test_case hyper_params.max_iter --choices all --random_test_times 20 --hp_values 1 2 3 4 10 15 -o ./.results/results_hyper_params_max_iter_all_0329.yaml
python ./main.py plot -f ./.results/results_hyper_params_max_iter_all_0329.yaml -x hyper_params.max_iter --labels all


python ./main.py plot -f ./.results/results_uav_num_all.json -x uav_num --labels all

# test uav num
python ./main.py test --test_case uav_num --choices centralized csci --random_test_times 5 --uav_nums 10 20 40 80 160 --task_nums 20 -o ./.results/results_uav_num_0516.yaml
python ./main.py plot -f ./.results/results_uav_num_0516.yaml -x uav_num --labels all


# 0518
## uav num
python ./main.py test --test_case uav_num --choices csci iros icra centralized distributed --random_test_times 20 --uav_nums 10 20 40 80 160 320 --task_nums 30 -o ./.results/results_uav_num_0518.yaml

python ./main.py plot -f  ./.results/results_uav_num_06.yaml -x uav_num --labels all --choices CSCI2024_Xue IROS2024_LiwangZhang ICRA2024_LiwangZhang Centralized Distributed --show
# --save_dir ./.results/uav_num_all_06/  

python ./main.py plot -f ./.results/results_uav_num_all_0518.yaml -x uav_num --labels all --choices CSCI2024_Xue IROS2024_LiwangZhang ICRA2024_LiwangZhang Centralized Distributed --save_dir ./.results/uav_num_all_0518/  --show

## task num
python ./main.py test --test_case task_num --choices csci iros icra centralized distributed --random_test_times 20 --uav_nums 40  --task_nums 10 20 40 80 160 -o ./.results/results_task_num_0518.yaml

python ./main.py plot -f ./.results/results_task_num_0518.yaml -x task_num --labels all --choices CSCI2024_Xue IROS2024_LiwangZhang ICRA2024_LiwangZhang Centralized Distributed --save_dir ./.results/task_num_all_0518/  --show

## hyper params
### resource contribution weight
python ./main.py test --test_case hyper_params.resource_contribution_weight --choices all --random_test_times 20 --hp_values -64.0 -10.0 -1.0 0.0 1.0 4.0 8.0 16.0 64.0 -o ./.results/results_hyper_params.resource_contribution_weight_all_0519.yaml

python ./main.py test --test_case hyper_params.resource_contribution_weight --choices all --random_test_times 10 --hp_values -10.0 -1.0 0.0 4.0 8.0 16.0 -o ./.results/results_hyper_params.resource_contribution_weight_all_06.yaml

python ./main.py plot -f ./.results/results_hyper_params.resource_contribution_weight_all_06.yaml -x hyper_params.resource_contribution_weight --labels all  --show 

python ./main.py plot -f ./.results/results_hyper_params.resource_contribution_weight_all_0519.yaml -x hyper_params.resource_contribution_weight --labels all --choices CSCI2024_Xue IROS2024_LiwangZhang ICRA2024_LiwangZhang Centralized Distributed  --show --save_dir ./.results/results_hyper_params.resource_contribution_weight_all_0519

### path cost weight
python ./main.py test --test_case hyper_params.path_cost_weight --choices all --random_test_times 20 --hp_values -10.0 -1.0 0.0 1.0 4.0 8.0 16.0 64.0 -o ./.results/results_hyper_params.path_cost_weight_all_0519.yaml

python ./main.py plot -f ./.results/results_hyper_params.path_cost_weight_all_0519.yaml -x hyper_params.path_cost_weight --labels all --show --save_dir ./.results/results_hyper_params.path_cost_weight_all_0519

### threat loss weight
python ./main.py test --test_case hyper_params.threat_loss_weight --choices all --random_test_times 20 --hp_values -10.0 -1.0 0.0 1.0 4.0 8.0 16.0 64.0 -o ./.results/results_hyper_params.threat_loss_weight_all_0519.yaml

python ./main.py plot -f ./.results/results_hyper_params.threat_loss_weight_all_0519.yaml -x hyper_params.threat_loss_weight --labels all --show --save_dir ./.results/results_hyper_params.threat_loss_weight_all_0519

### resource waste weight
python ./main.py test --test_case hyper_params.resource_waste_weight --choices centralized distributed --random_test_times 20 --hp_values -10.0 -1.0 0.0 1.0 4.0 8.0 16.0 64.0 -o ./.results/results_hyper_params.resource_waste_weight_all_0519.yaml

python ./main.py plot -f ./.results/results_hyper_params.resource_waste_weight_all_0519.yaml -x hyper_params.resource_waste_weight --labels all --show --save_dir ./.results/results_hyper_params.resource_waste_weight_all_0519

## example [TODO]

## Ablation Study

python ./main.py test --test_case uav_num --choices centralized distributed centralized_random_init distributed_random_init --random_test_times 10 --uav_nums 10 20 40 80 --task_nums 30 -o ./.results/results_uav_num_ablation_06.yaml

python ./main.py plot -f ./.results/results_uav_num_ablation_06.yaml -x uav_num --labels all --show

### uav num
python ./main.py test --test_case uav_num --choices centralized distributed centralized_random_init distributed_random_init --random_test_times 20 --uav_nums 10 20 40 80 160 320 --task_nums 30 -o ./.results/results_uav_num_ablation_0519.yaml

python ./main.py plot -f ./.results/results_uav_num_ablation_0519.yaml -x uav_num --labels all --save_dir ./.results/uav_num_all_ablation_0519/  --show

### task num
python ./main.py test --test_case task_num --choices centralized distributed centralized_random_init distributed_random_init --random_test_times 20 --uav_nums 40  --task_nums 10 20 40 80 160 -o ./.results/results_task_num_ablation_0519.yaml

python ./main.py plot -f ./.results/results_task_num_ablation_0519.yaml -x task_num --labels all --save_dir ./.results/task_num_all_ablation_0519/  --show

### preference
python ./main.py test --test_case uav_num --choices centralized centralized_selfish centralized_pareto distributed distributed_selfish distributed_pareto --random_test_times 20 --uav_nums 10 20 40 80 --task_nums 30 -o ./.results/results_uav_num_preference_0519.yaml

python ./main.py plot -f ./.results/results_uav_num_preference_0519.yaml -x uav_num --labels all  --save_dir ./.results/uav_num_all_preference_0519/
--show

## hyper params
### resource contribution weight
python ./main.py test --test_case uav_num --choices  centralized_selfish centralized_pareto  distributed_selfish distributed_pareto --random_test_times 1 --uav_nums 80  --task_nums 30
$$
## exchange
python ./main.py test --test_case uav_num \
--choices centralized centralized_exchange \
--random_test_times 10 \
--uav_nums 10 20 40 80 120 160 320 --task_nums 100 \
-o ./.results/results_uav_num_exchange_0526.yaml

python ./main.py plot -f ./.results/results_uav_num_exchange_0526.yaml -x uav_num --labels all --save_dir ./.results/uav_num_all_exchange_0526/ --show

all_labels = [
    "elapsed_time",
    "completion_rate",
    "resource_use_rate",
    "total_distance",
    "total_energy",
    "total_exploss",
]

About

A repo for swarm uavs task assignment simulation and algorithm implementation.

Resources

Readme
Activity

Stars

6 stars

Watchers

1 watching

Forks

1 fork
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages