Accurate trajectory prediction of surrounding road users is the fundamental input for motion planning,which enables safe autonomous driving on public roads.In this paper,a safe motion planning approach is proposed bas...Accurate trajectory prediction of surrounding road users is the fundamental input for motion planning,which enables safe autonomous driving on public roads.In this paper,a safe motion planning approach is proposed based on the deep learning-based trajectory prediction method.To begin with,a trajectory prediction model is established based on the graph neural network(GNN)that is trained utilizing the INTERACTION dataset.Then,the validated trajectory prediction model is used to predict the future trajectories of surrounding road users,including pedestrians and vehicles.In addition,a GNN prediction model-enabled motion planner is developed based on the model predictive control technique.Furthermore,two driving scenarios are extracted from the INTERACTION dataset to validate and evaluate the effectiveness of the proposed motion planning approach,i.e.,merging and roundabout scenarios.The results demonstrate that the proposed method can lower the risk and improve driving safety compared with the baseline method.展开更多
In order to make motion planning fitting practice,many characteristic of CNC trajectory motion are discussed, such as the geometric function,the motion and the time.It is found that the relation between orbit function...In order to make motion planning fitting practice,many characteristic of CNC trajectory motion are discussed, such as the geometric function,the motion and the time.It is found that the relation between orbit function and motional parame- ter,so the differential equation about the trajectory motion be set-up by the goal of trajectory motion.The actual motion process is defined as reference time to link planning and practice.Present a new movement planning method based on self-defining time.At rest state,the differential simultaneous equation can be calculated according geometric characteristic analysis,it can be get that simple function consisted of coordinate and reference time variants.At motive state,dynamic parameter can be worked out accord- ing practical value of reference time,It is proved by experiment and simulation that it is a good way to control geometry and motion comprehensively,to reduce computation times and to increase the ability of environmental adaptation for path展开更多
This paper proposes a unified trajectory optimization approach that simultaneously optimizes the trajectory of the center of mass and footholds for legged locomotion.Based on a generic point-mass model,the approach is...This paper proposes a unified trajectory optimization approach that simultaneously optimizes the trajectory of the center of mass and footholds for legged locomotion.Based on a generic point-mass model,the approach is formulated as a nonlinear optimization problem,incorporating constraints such as robot kinematics,dynamics,ground reaction forces,obstacles,and target location.The unified optimization approach can be applied to both long-term motion planning and the reactive online planning through the use of model predictive control,and it incorporates vector field guidance to converge to the long-term planned motion.The effectiveness of the approach is demonstrated through simulations and physical experiments,showing its ability to generate a variety of walking and jumping gaits,as well as transitions between them,and to perform reactive walking in obstructed environments.展开更多
基金Supported by National Natural Science Foundation of China(Grant Nos.52222215,52072051)Chongqing Municipal Natural Science Foundation of China(Grant No.CSTB2023NSCQ-JQX0003).
文摘Accurate trajectory prediction of surrounding road users is the fundamental input for motion planning,which enables safe autonomous driving on public roads.In this paper,a safe motion planning approach is proposed based on the deep learning-based trajectory prediction method.To begin with,a trajectory prediction model is established based on the graph neural network(GNN)that is trained utilizing the INTERACTION dataset.Then,the validated trajectory prediction model is used to predict the future trajectories of surrounding road users,including pedestrians and vehicles.In addition,a GNN prediction model-enabled motion planner is developed based on the model predictive control technique.Furthermore,two driving scenarios are extracted from the INTERACTION dataset to validate and evaluate the effectiveness of the proposed motion planning approach,i.e.,merging and roundabout scenarios.The results demonstrate that the proposed method can lower the risk and improve driving safety compared with the baseline method.
基金Supported by the Natural Science Foundation of Education Committee of Sichuan Province(2004A163)
文摘In order to make motion planning fitting practice,many characteristic of CNC trajectory motion are discussed, such as the geometric function,the motion and the time.It is found that the relation between orbit function and motional parame- ter,so the differential equation about the trajectory motion be set-up by the goal of trajectory motion.The actual motion process is defined as reference time to link planning and practice.Present a new movement planning method based on self-defining time.At rest state,the differential simultaneous equation can be calculated according geometric characteristic analysis,it can be get that simple function consisted of coordinate and reference time variants.At motive state,dynamic parameter can be worked out accord- ing practical value of reference time,It is proved by experiment and simulation that it is a good way to control geometry and motion comprehensively,to reduce computation times and to increase the ability of environmental adaptation for path
基金supported by the Natural Science Foundation of Hebei Province of China(no.E2022203095)Cultivation Project for Basic Research and Innovation of Yanshan University(no.2021LGQN004)National Natural Science Foundation of China(no.51905465 and No.52122503).
文摘This paper proposes a unified trajectory optimization approach that simultaneously optimizes the trajectory of the center of mass and footholds for legged locomotion.Based on a generic point-mass model,the approach is formulated as a nonlinear optimization problem,incorporating constraints such as robot kinematics,dynamics,ground reaction forces,obstacles,and target location.The unified optimization approach can be applied to both long-term motion planning and the reactive online planning through the use of model predictive control,and it incorporates vector field guidance to converge to the long-term planned motion.The effectiveness of the approach is demonstrated through simulations and physical experiments,showing its ability to generate a variety of walking and jumping gaits,as well as transitions between them,and to perform reactive walking in obstructed environments.
