Motion planning is one of the most significant technologies for autonomous driving. To make motion planning models able to learn from the environment and to deal with emergency situations, a new motion planning framew...Motion planning is one of the most significant technologies for autonomous driving. To make motion planning models able to learn from the environment and to deal with emergency situations, a new motion planning framework called as"parallel planning" is proposed in this paper. In order to generate sufficient and various training samples, artificial traffic scenes are firstly constructed based on the knowledge from the reality.A deep planning model which combines a convolutional neural network(CNN) with the Long Short-Term Memory module(LSTM) is developed to make planning decisions in an end-toend mode. This model can learn from both real and artificial traffic scenes and imitate the driving style of human drivers.Moreover, a parallel deep reinforcement learning approach is also presented to improve the robustness of planning model and reduce the error rate. To handle emergency situations, a hybrid generative model including a variational auto-encoder(VAE) and a generative adversarial network(GAN) is utilized to learn from virtual emergencies generated in artificial traffic scenes. While an autonomous vehicle is moving, the hybrid generative model generates multiple video clips in parallel, which correspond to different potential emergency scenarios. Simultaneously, the deep planning model makes planning decisions for both virtual and current real scenes. The final planning decision is determined by analysis of real observations. Leveraging the parallel planning approach, the planner is able to make rational decisions without heavy calculation burden when an emergency occurs.展开更多
Unmanned Aerial Vehicles(UAVs)are useful in dangerous and dynamic tasks such as search-and-rescue,forest surveillance,and anti-terrorist operations.These tasks can be solved better through the collaboration of multipl...Unmanned Aerial Vehicles(UAVs)are useful in dangerous and dynamic tasks such as search-and-rescue,forest surveillance,and anti-terrorist operations.These tasks can be solved better through the collaboration of multiple UAVs under human supervision.However,it is still difficult for human to monitor,understand,predict and control the behaviors of the UAVs due to the task complexity as well as the black-box machine learning and planning algorithms being used.In this paper,the coactive design method is adopted to analyze the cognitive capabilities required for the tasks and design the interdependencies among the heterogeneous teammates of UAVs or human for coherent collaboration.Then,an agent-based task planner is proposed to automatically decompose a complex task into a sequence of explainable subtasks under constrains of resources,execution time,social rules and costs.Besides,a deep reinforcement learning approach is designed for the UAVs to learn optimal policies of a flocking behavior and a path planner that are easy for the human operator to understand and control.Finally,a mixed-initiative action selection mechanism is used to evaluate the learned policies as well as the human’s decisions.Experimental results demonstrate the effectiveness of the proposed methods.展开更多
基金supported in part by the National Natural Science Foundation of China (61773414,61806076)Hubei Provincial Natural Science Foundation of China (2018CFB158)
文摘Motion planning is one of the most significant technologies for autonomous driving. To make motion planning models able to learn from the environment and to deal with emergency situations, a new motion planning framework called as"parallel planning" is proposed in this paper. In order to generate sufficient and various training samples, artificial traffic scenes are firstly constructed based on the knowledge from the reality.A deep planning model which combines a convolutional neural network(CNN) with the Long Short-Term Memory module(LSTM) is developed to make planning decisions in an end-toend mode. This model can learn from both real and artificial traffic scenes and imitate the driving style of human drivers.Moreover, a parallel deep reinforcement learning approach is also presented to improve the robustness of planning model and reduce the error rate. To handle emergency situations, a hybrid generative model including a variational auto-encoder(VAE) and a generative adversarial network(GAN) is utilized to learn from virtual emergencies generated in artificial traffic scenes. While an autonomous vehicle is moving, the hybrid generative model generates multiple video clips in parallel, which correspond to different potential emergency scenarios. Simultaneously, the deep planning model makes planning decisions for both virtual and current real scenes. The final planning decision is determined by analysis of real observations. Leveraging the parallel planning approach, the planner is able to make rational decisions without heavy calculation burden when an emergency occurs.
基金co-supported by the National Natural Science Foundation of China(Nos.61906203,61876187)the National Key Laboratory of Science and Technology on UAV,Northwestern Polytechnical University,China(No.614230110080817)。
文摘Unmanned Aerial Vehicles(UAVs)are useful in dangerous and dynamic tasks such as search-and-rescue,forest surveillance,and anti-terrorist operations.These tasks can be solved better through the collaboration of multiple UAVs under human supervision.However,it is still difficult for human to monitor,understand,predict and control the behaviors of the UAVs due to the task complexity as well as the black-box machine learning and planning algorithms being used.In this paper,the coactive design method is adopted to analyze the cognitive capabilities required for the tasks and design the interdependencies among the heterogeneous teammates of UAVs or human for coherent collaboration.Then,an agent-based task planner is proposed to automatically decompose a complex task into a sequence of explainable subtasks under constrains of resources,execution time,social rules and costs.Besides,a deep reinforcement learning approach is designed for the UAVs to learn optimal policies of a flocking behavior and a path planner that are easy for the human operator to understand and control.Finally,a mixed-initiative action selection mechanism is used to evaluate the learned policies as well as the human’s decisions.Experimental results demonstrate the effectiveness of the proposed methods.
