Abstract:Trajectory prediction is a crucial aspect of understanding human behaviors. Researchers have made efforts to represent socially interactive behaviors among pedestrians and utilize various networks to enhance prediction capability. Unfortunately, they still face challenges not only in fully explaining and measuring how these interactive behaviors work to modify trajectories but also in modeling pedestrians' preferences to plan or participate in social interactions in response to the changeable physical environments as extra conditions. This manuscript mainly focuses on the above explainability and conditionality requirements for trajectory prediction networks. Inspired by marine animals perceiving other companions and the environment underwater by echolocation, this work constructs an angle-based conditioned social interaction representation SocialCircle+ to represent the socially interactive context and its corresponding conditions. It employs a social branch and a conditional branch to describe how pedestrians are positioned in prediction scenes socially and physically in angle-based-cyclic-sequence forms. Then, adaptive fusion is applied to fuse the above conditional clues onto the social ones to learn the final interaction representation. Experiments demonstrate the superiority of SocialCircle+ with different trajectory prediction backbones. Moreover, counterfactual interventions have been made to simultaneously verify the modeling capacity of causalities among interactive variables and the conditioning capability.

What problem does this paper attempt to address?

### What problems does this paper attempt to solve? This paper mainly focuses on two core challenges in pedestrian trajectory prediction: **Explainability** and **Conditionality**. Specifically: 1. **Explainability**: - Most of the current trajectory prediction models are data - driven, and it is difficult to explain how these models simulate social interactions and their impact on trajectories. Especially in complex multi - agent systems, social interactions are usually accompanied by causal relationships, and existing models have difficulty in capturing and explaining these causal relationships. - The paper points out that although some studies have introduced causal analysis methods, few studies have analyzed from the perspective of causal analysis during the verification process. Therefore, constructing a model that can explain the causal relationships of social interactions is an important challenge. 2. **Conditionality**: - Most of the existing trajectory prediction studies focus on the impact of social interactions or scene constraints on trajectories, but ignore the conditional effect of the physical environment on social interactions. For example, in different physical environments (such as sidewalks, crossroads, basketball courts), the ways of social interactions will be different. - The influence of the physical environment on social interactions is conditional, that is, social interactions are restricted by environmental factors. However, most existing methods do not fully consider this point when modeling, resulting in poor adaptability of the model in different scenarios. To solve the above problems, the paper proposes the **SocialCircle+** representation method. By modeling social interactions based on angles to simulate the sonar positioning process, the explainability and conditionality of the trajectory prediction model are improved. Specific contributions include: - **Angle - based cyclic interaction modeling strategy**: Combine all social interaction behaviors with environmental conditions, and model based on angles to form a head - to - tail connected cyclic sequence representation. - **Three angle - based PhysicalCircle meta - components**: Used to represent the interaction conditions of the physical environment around the target. - **SocialCircle+ representation method**: Through the partition - adaptive fusion strategy, integrate the physical environment conditions into the social interaction representation, so that the prediction network can learn and represent social interactions while considering the physical environment. Through these improvements, the paper aims to improve the performance of the trajectory prediction model in complex and changeable environments and provide better causal explanation capabilities.

SocialCircle+: Learning the Angle-based Conditioned Interaction Representation for Pedestrian Trajectory Prediction

SocialCVAE: Predicting Pedestrian Trajectory via Interaction Conditioned Latents

CR-LSTM: Collision-prior Guided Social Refinement for Pedestrian Trajectory Prediction

Who Walks With You Matters: Perceiving Social Interactions with Groups for Pedestrian Trajectory Prediction

SSAGCN: Social Soft Attention Graph Convolution Network for Pedestrian Trajectory Prediction

Modeling interpretable social interactions for pedestrian trajectory

Spatio-Temporal Interaction Aware and Trajectory Distribution Aware Graph Convolution Network for Pedestrian Multimodal Trajectory Prediction

Crossmodal Transformer Based Generative Framework for Pedestrian Trajectory Prediction

SGCN:Sparse Graph Convolution Network for Pedestrian Trajectory Prediction

Modeling social interaction and intention for pedestrian trajectory prediction

Stochastic trajectory prediction with social graph network

Learning Sparse Interaction Graphs of Partially Detected Pedestrians for Trajectory Prediction

Socially-Informed Reconstruction for Pedestrian Trajectory Forecasting

Multi-Relational Pedestrian Trajectory Prediction in Complex Scenes.

CSR: Cascade Conditional Variational AutoEncoder with Social-aware Regression for Pedestrian Trajectory Prediction

Pedestrian Trajectory Prediction Based on Social Interactions Learning With Random Weights

A Set of Control Points Conditioned Pedestrian Trajectory Prediction

Social-Aware Pedestrian Trajectory Prediction via States Refinement LSTM

A Unified Environmental Network for Pedestrian Trajectory Prediction

Socially-Aware Graph Convolutional Network for Human Trajectory Prediction

SCSG Attention: A Self-centered Star Graph with Attention for Pedestrian Trajectory Prediction