BiGRU based online multi-modal driving maneuvers and trajectory prediction

2021 ◽  
pp. 095440702110143
Author(s):  
Yongshuai Zhi ◽  
Zhipeng Bao ◽  
Sumin Zhang ◽  
Rui He
Keyword(s):  
Autonomous Vehicles ◽  
Intelligent Systems ◽  
Predictive Distribution ◽  
Behavior Prediction ◽  
Trajectory Prediction ◽  
Prediction Horizon ◽  
Vehicle Motion ◽  
Quality Decision ◽  
Gated Recurrent Unit ◽  
Complete History

Accurately predicting maneuvers and trajectory of vehicles are essential prerequisites for intelligent systems such as autonomous vehicles to achieve safe and high-quality decision making and motion planning. Motions of each vehicle in a scene is governed by the traffic context, that is, the motion and relative spatial positions of neighboring vehicles, and is also affected by its motion inertia, that is, the trajectory history. In this paper, we propose a novel scheme based on Bidirectional Gated Recurrent Unit (BiGRU) to conduct online multi-modal driving maneuvers and trajectory prediction. The motivation for this BiGRU based method relies on its enhanced prediction accuracy and computational efficiency in outputting the predicted results within the limited prediction horizon. We utilize a BiGRU to extract the complete history and future information of every point in the trajectory history sequence, apply dilated convolutional social (DCS) for learning interdependencies in vehicle motion, and subsequently use a GRU decoder model to make predictions. Additionally, our model simultaneously outputs a multi-modal predictive distribution over future trajectory and vehicle’s behavior prediction results. We evaluate our model using the publicly available NGSIM US-101and I-80 datasets. Our results show improvements over the state-of-the-art in terms of Root Mean Square Error (RMSE) values and Negative Log-Likelihoods (NLL). We also present a qualitative analysis of the model’s predicted maneuvers and multi-model trajectories for various traffic scenarios.

Perspectives on Ethics of AI

2020 ◽  
pp. 419-441
Author(s):  
Benjamin Kuipers
Keyword(s):  
Autonomous Vehicles ◽  
Intelligent Systems ◽  
Feedback Mechanism ◽  
Computational Approach ◽  
Human Society ◽  
Decision Point ◽  
Free Riders ◽  
Ethical Knowledge ◽  
Distributed Intelligent Systems ◽  
Lesser Evil

This chapter describes a computational view of the function of ethics in human society and discusses its application to three diverse examples. First, autonomous vehicles are individually embodied intelligent systems that act as members of society. The ethical knowledge needed by such an agent is not how to choose the lesser evil when confronted by a Deadly Dilemma, but how to recognize the upstream decision point that makes it possible to avoid the Deadly Dilemma entirely. Second, disembodied distributed intelligent systems like Google and Facebook provide valuable services while collecting, aggregating, and correlating vast amounts of information about individual users. With inadequate controls, these corporate systems can invade privacy and do substantial damage through either correct or incorrect inferences. Third, acceptance of the legitimacy of the society by its individual members depends on a general perception of fairness. Rage about unfairness can be directed at individual free-riders or at systematic inequality across the society. Ultimately, the promise of a computational approach to ethical knowledge is not simply ethics for computational devices such as robots. It also promises to help people understand the pragmatic value of ethics as a feedback mechanism that helps intelligent creatures, human and nonhuman, live together in thriving societies.

scholarly journals Design of AI-based gamification platform for effective educational service using child behavior prediction/change

2021 ◽  
Vol 12 (5) ◽  
pp. 286-292
Author(s):  
Hyeon Woo Nam
Keyword(s):  
Artificial Intelligence ◽  
Autonomous Vehicles ◽  
Industrial Revolution ◽  
Teaching Method ◽  
Behavior Prediction ◽  
Interactive Communication ◽  
Service Platform ◽  
Learning Attitudes ◽  
Education Industry ◽  
Artificial Intelligence Technology

Due to the advancement of advanced technologies such as artificial intelligence, robots, autonomous vehicles, healthcare, virtual reality, augmented reality, etc. and the popularization of smartphones, it stimulates customer interest and leads voluntary participation in order to maximize interactive communication in all industries The gamification strategy incorporating games began to emerge. A representative field that generates results by easily introducing such a gamification strategy is the education industry that seeks to improve the educational effect by utilizing the elements of corporate marketing strategies and games such as challenge, competition, achievement, and reward. Recently, gamification research is being conducted to effectively apply AI and big data, the core technologies of the 4th industrial revolution in all industries. Gamification is actively forming markets in Europe and the US, and it can increase customer loyalty and productivity by applying various roles applied to games in other industries as well as serious games. The purpose of this study is to design and implement a gamification service platform based on artificial intelligence technology and operate the implemented system to expand the area where the gamification service applied to the existing marketing and consulting fields can be used. The designed gamification service platform can be applied to education services that increase learning efficiency by analyzing the predicted learning attitudes of trainees, and through successful research cases, it will be able to provide immersion effect to trainees and teaching method research to educators.

scholarly journals Road-Aware Trajectory Prediction for Autonomous Driving on Highways

Sensors ◽  
2020 ◽  
Vol 20 (17) ◽  
pp. 4703
Author(s):  
Yookhyun Yoon ◽  
Taeyeon Kim ◽  
Ho Lee ◽  
Jahnghyon Park
Keyword(s):  
Deep Learning ◽  
Autonomous Vehicles ◽  
Prediction Method ◽  
Autonomous Driving ◽  
Structural Constraints ◽  
High Definition ◽  
Trajectory Prediction ◽  
The Road ◽  
Road Geometry ◽  
Efficient Learning

For driving safely and comfortably, the long-term trajectory prediction of surrounding vehicles is essential for autonomous vehicles. For handling the uncertain nature of trajectory prediction, deep-learning-based approaches have been proposed previously. An on-road vehicle must obey road geometry, i.e., it should run within the constraint of the road shape. Herein, we present a novel road-aware trajectory prediction method which leverages the use of high-definition maps with a deep learning network. We developed a data-efficient learning framework for the trajectory prediction network in the curvilinear coordinate system of the road and a lane assignment for the surrounding vehicles. Then, we proposed a novel output-constrained sequence-to-sequence trajectory prediction network to incorporate the structural constraints of the road. Our method uses these structural constraints as prior knowledge for the prediction network. It is not only used as an input to the trajectory prediction network, but is also included in the constrained loss function of the maneuver recognition network. Accordingly, the proposed method can predict a feasible and realistic intention of the driver and trajectory. Our method has been evaluated using a real traffic dataset, and the results thus obtained show that it is data-efficient and can predict reasonable trajectories at merging sections.

scholarly journals TrafficPredict: Trajectory Prediction for Heterogeneous Traffic-Agents

2019 ◽  
Vol 33 ◽  
pp. 6120-6127 ◽  
Author(s):  
Yuexin Ma ◽  
Xinge Zhu ◽  
Sibo Zhang ◽  
Ruigang Yang ◽  
Wenping Wang ◽  
...  
Keyword(s):  
Autonomous Vehicles ◽  
Short Term Memory ◽  
Autonomous Vehicle ◽  
Large City ◽  
Urban Traffic ◽  
Heterogeneous Traffic ◽  
Prediction Algorithm ◽  
Trajectory Prediction ◽  
Long Short Term Memory ◽  
Traffic Agents

To safely and efficiently navigate in complex urban traffic, autonomous vehicles must make responsible predictions in relation to surrounding traffic-agents (vehicles, bicycles, pedestrians, etc.). A challenging and critical task is to explore the movement patterns of different traffic-agents and predict their future trajectories accurately to help the autonomous vehicle make reasonable navigation decision. To solve this problem, we propose a long short-term memory-based (LSTM-based) realtime traffic prediction algorithm, TrafficPredict. Our approach uses an instance layer to learn instances’ movements and interactions and has a category layer to learn the similarities of instances belonging to the same type to refine the prediction. In order to evaluate its performance, we collected trajectory datasets in a large city consisting of varying conditions and traffic densities. The dataset includes many challenging scenarios where vehicles, bicycles, and pedestrians move among one another. We evaluate the performance of TrafficPredict on our new dataset and highlight its higher accuracy for trajectory prediction by comparing with prior prediction methods.

scholarly journals Coupling Intent and Action for Pedestrian Crossing Behavior Prediction

2021 ◽  
Author(s):  
Yu Yao ◽  
Ella Atkins ◽  
Matthew Johnson-Roberson ◽  
Ram Vasudevan ◽  
Xiaoxiao Du
Keyword(s):  
Traffic Safety ◽  
Autonomous Vehicles ◽  
State Of The Art ◽  
Semantic Interpretation ◽  
Behavior Prediction ◽  
Probabilistic Representation ◽  
Class Actions ◽  
The Future ◽  
Naturalistic Driving ◽  
Pedestrian Crossing

Accurate prediction of pedestrian crossing behaviors by autonomous vehicles can significantly improve traffic safety. Existing approaches often model pedestrian behaviors using trajectories or poses but do not offer a deeper semantic interpretation of a person's actions or how actions influence a pedestrian's intention to cross in the future. In this work, we follow the neuroscience and psychological literature to define pedestrian crossing behavior as a combination of an unobserved inner will (a probabilistic representation of binary intent of crossing vs. not crossing) and a set of multi-class actions (e.g., walking, standing, etc.). Intent generates actions, and the future actions in turn reflect the intent. We present a novel multi-task network that predicts future pedestrian actions and uses predicted future action as a prior to detect the present intent and action of the pedestrian. We also designed an attention relation network to incorporate external environmental contexts thus further improve intent and action detection performance. We evaluated our approach on two naturalistic driving datasets, PIE and JAAD, and extensive experiments show significantly improved and more explainable results for both intent detection and action prediction over state-of-the-art approaches. Our code is available at: https://github.com/umautobots/pedestrian_intent_action_detection

Sign in / Sign up

Export Citation Format

Share Document