Abstract:Recently, the application of autonomous driving in open-pit mining has garnered increasing attention for achieving safe and efficient mineral transportation. Compared to urban structured roads, unstructured roads in mining sites have uneven boundaries and lack clearly defined lane markings. This leads to a lack of sufficient constraint information for predicting the trajectories of other human-driven vehicles, resulting in higher uncertainty in trajectory prediction problems. A method is proposed to predict multiple possible trajectories and their probabilities of the target vehicle. The surrounding environment and historical trajectories of the target vehicle are encoded as a rasterized image, which is used as input to our deep convolutional network to predict the target vehicle's multiple possible trajectories. The method underwent offline testing on a dataset specifically designed for autonomous driving scenarios in open-pit mining and was compared and evaluated against physics-based method. The open-source code and data are available at <a class="link-external link-https" href="https://github.com/LLsxyc/mine_motion_prediction.git" rel="external noopener nofollow">this https URL</a>

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to perform multi - modal trajectory prediction on unstructured roads in open - pit mining areas in order to achieve the safe and efficient operation of self - driving vehicles. Specifically, compared with the structured roads in cities, the road boundaries in open - pit mining areas are irregular and lack clear lane markings, which leads to insufficient constraint information for predicting the trajectories of other human - driven vehicles and increases the uncertainty of trajectory prediction. ### Main problems: 1. **Unstructured road environment**: Roads in open - pit mining areas do not have clear lane lines, traffic lights and other facilities, making the behaviors of vehicles (such as turning left, turning right or going straight) more diverse and difficult to predict. 2. **High uncertainty in trajectory prediction**: Due to the lack of sufficient constraint information, it becomes more difficult to predict the future trajectories of other human - driven vehicles, especially in complex road conditions. 3. **Limitations of existing methods**: Traditional physics - based methods perform poorly in long - term trajectory prediction, and existing learning methods mainly focus on structured road networks (such as urban roads and highways) and have limited applications in unstructured road scenarios. ### Solutions: The paper proposes a method based on deep convolutional neural network (CNN) to solve the above problems through the following steps: - **Encoding historical trajectories and surrounding environment**: Encode the historical trajectory of the target vehicle and the surrounding environment into rasterized images as the input of the CNN. - **Multi - modal trajectory generation**: Use the CNN model to predict multiple possible trajectories of the target vehicle and their probabilities. - **Loss function design**: Design a loss function specifically for multi - modal trajectory prediction to improve the accuracy and robustness of prediction. Through this method, the paper aims to improve the safety and efficiency of self - driving vehicles on unstructured roads in open - pit mining areas, especially when sharing the road with other human - driven vehicles. ### Formula summary: - **Average Displacement Error (ADE)**: \[ L_{h}=\frac{1}{H} \sum_{h = 1}^{H}\left\|x_{T + h}-\hat{x}_{T + h}\right\|_{2} \] - **Mixture of Experts (ME) loss**: \[ L_{ME}=\sum_{m = 1}^{M}p_{m} \cdot\left\|x_{T}^{m}-\hat{x}_{T}\right\|_{2} \] - **Final loss function**: \[ L_{final}=\alpha \cdot L_{class}+(1-\alpha) \cdot L_{reg} \] where \(L_{class}\) is the classification cross - entropy loss, \(L_{reg}\) is the regression loss, and \(\alpha\) is the balancing parameter. Through these methods and formulas, the paper effectively solves the challenges of multi - modal trajectory prediction on unstructured roads.

Multimodal Trajectory Prediction for Autonomous Driving on Unstructured Roads using Deep Convolutional Network

Multimodal Trajectory Predictions for Autonomous Driving using Deep Convolutional Networks

Enhanced Multimodal Trajectory Prediction for Autonomous Vehicles Using Advanced Diffusion Model Techniques

Multi-modal Trajectory Prediction for Autonomous Driving with Semantic Map and Dynamic Graph Attention Network

Graph-Based Interaction-Aware Multimodal 2D Vehicle Trajectory Prediction using Diffusion Graph Convolutional Networks

MFTraj: Map-Free, Behavior-Driven Trajectory Prediction for Autonomous Driving

An Improved Multimodal Trajectory Prediction Method Based on Deep Inverse Reinforcement Learning

Probabilistic multi-modal expected trajectory prediction based on LSTM for autonomous driving

Multiple Contextual Cues Integrated Trajectory Prediction for Autonomous Driving

A Cognitive-Based Trajectory Prediction Approach for Autonomous Driving

Characterized Diffusion and Spatial-Temporal Interaction Network for Trajectory Prediction in Autonomous Driving

Trajectory Prediction with Attention-Based Spatial–Temporal Graph Convolutional Networks for Autonomous Driving

Multi-Modal Vehicle Trajectory Prediction by Collaborative Learning of Lane Orientation, Vehicle Interaction, and Intention

Context‐aware trajectory prediction for autonomous driving in heterogeneous environments

A Multi-Modal States based Vehicle Descriptor and Dilated Convolutional Social Pooling for Vehicle Trajectory Prediction

Multimodal Trajectory Predictions for Autonomous Driving without a Detailed Prior Map

Path-based Multimodal Trajectories Prediction

Vehicle Trajectory Prediction with Multimodal and Dynamics-Aware Interaction Neural Networks

Shared Cross-Modal Trajectory Prediction for Autonomous Driving

Trajectory Prediction for Autonomous Driving with Topometric Map

A Cognitive-Driven Trajectory Prediction Model for Autonomous Driving in Mixed Autonomy Environment