MineSim

MineSim Simulator

MineSim, an open-source, scenario-based simulation test system specifically developed for planning tasks in open-pit mines. MineSim consists of three main components: the Simulation Engine, the Metric System, and the Scenario Visualization tool. The Simulation Engine serves as the core of MineSim.It loads and parses scenarios from the scenario library via the Environment Manager, then implements closed-loop simulation tests for autonomous mining trucks through the sequential execution of the Planning Algorithm, Ego Update Model, and Agent Update Policy.

The Ego Update Model is responsible for updating the state of the ego vehicle in the simulation. It receives actions output by the tested algorithm, represented as [acceleration, steering angle], and generates the ego vehicle's updated state for the next step. The Agent Update Policy uses multiple agent objects and environmental states as inputs to determine the appropriate actions and next state for controllable agents within the scenario. The Metric System enables users to create a customized evaluation framework focused on specific performance after a set of scenario tests. It includes four main categories: safety, efficiency, smoothness, and task completion. The Scenario Visualization tool provides both 2D top-down and 3D views of the test scenarios, helping users better understand and demonstrate the proposed algorithms' performance.

Scenario 2D Visualizer

Scenario ID:dapai_intersection_1_3_4. Planning Method: Sampling Planner using Predefined Maneuver Modes (SPPMM).

Scenario 3D Visualizer

Scenario ID:dapai_intersection_1_3_4. Planning Method: SPPMM. An example of a collision occurring.

Dynamic obstacle scenario library and benchmark

This library focuses on testing obstacle avoidance in mixed traffic at mining intersections. Its scenarios feature unstructured intersections with varying slopes and irregular shapes, where autonomous mining trucks interact with other vehicles. The main goal is to assess the trucks' smooth obstacle avoidance. Task settings and benchmark algorithms are detailed in Code Repository: MineSim-Dynamic.

Static obstacle scenario library and benchmark

This library is designed to test obstacle avoidance with static obstacles on mining roads. It records scenarios where the trucks face static obstacles of various sizes on roads with different curves and widths. The focus is on assessing the ability of autonomous mining trucks to plan smooth, efficient trajectories. Detailed task settings and benchmark algorithms are provided in Code Repository: MineSim-Static.

Scenario visualization tool

MineSim provides two Scenario Visualization Tools: the 2D Visualizer and the 3D Visualizer.The 2D Visualizer, developed using the Python matplotlib, renders top-down views of each frame of a single scenario based on the simulation log file. The 3D Visualizer, developed using ROS, renders 3D views of each frame in a scenario based on the simulation log file (see Code Repository: MineSim-3DVisualTool ).