Pedestrian trajectory prediction is crucial for enabling dynamic obstacle avoidance in social robots. Variational Autoencoders (VAE) have shown potential in predicting multi-modal distributions of future pedestrian trajectories. However, standards VAE struggle to generate accurate future trajectories, and existing prediction methods often overlook the relationships between pedestrian groups. This paper introduces a novel prediction model, called the Learning Autoencoder Diffusion Model (LADM) of Pedestrian Group Relationships for Multimodal Trajectory Prediction, which takes into account pedestrian group relationships, enhancing the accuracy of multimodal distribution trajectory prediction. In the LADM framework, each pedestrian is assigned to their most probable group through a learning process, and the interaction relationships between pedestrians and groups are determined using a pedestrian-group interaction module. To improve the quality of generated future trajectory distributions, we propose the Autoencoder Diffusion Model, the VAE functions as a generator and a diffusion model acts as a refiner. We evaluate our proposed method on two public datasets (ETH and UCY) and compare it to state-of-the-art methods. Experimental results demonstrate that our approach outperforms existing methods in terms of Average Displacement Error (ADE) and Final Displacement Error (FDE) metrics.

Since Vision Transformers (ViTs) are introduced into computer vision, they have developed rapidly in a variety of visual tasks. Recently, they have been gradually applied to visual tracking. The Transformer can adaptively capture the global similarity comparisons of target objects and search regions, which has achieved competitive performance results. However, Transformer architectures often require a large amount of training data and computing resources, and lack prior knowledge of inductive biases that existed in images. The advantages of convolutional neural networks (CNNs) in extracting local similarities are not fully exploited. To resolve these problems, we propose a lightweight tracking architecture, combining CNN and Transformer in the feature fusion stage. Specifically, Local-Global Feature Interaction (LGFI) module and Feature Cross-Fusion (FCF) module are the key components in our approach. In the LGFI module, the proposed method includes a Transformer global information network and a Transformer-like CNN local information network for simultaneous global scope dependency establishment and local feature similarity enhancement, then aggregates their feature results together. In the FCF module, the proposed method includes a multi-head cross-attention and a convolutional feedforward network for feature fusion of templates and search regions. Finally, we use the classification and regression head to predict the exact location of the target. Extensive experiments demonstrate that, our method achieves better tracking performance than the baseline method, when both methods are trained with fewer data. Meanwhile, without any extra training data, the proposed method also obtains comparable results with other state-of-the-art trackers on six challenging benchmarks, including GOT-10k, LaSOT, TrackingNet, OTB100, UAV123, and NFS. Furthermore, our model is lightweight compared with the baseline method, with fewer parameters and lower FLOPs, while running at real-time speed.

Recently Transformer has been largely explored in object tracking, and shown state-of-the-art (SOTA) performance compared to convolutional neural networks (CNNs). Especially, single-object trackers based on pure Transformer and ”CNN+Transformer” frameworks have achieved great success in terms of accuracy and speed. However, most methods do not fully exploit the temporal and spatial information of targets. Furthermore, the potential of trackers for spatial information interaction and propagation between the search area and templates remains under-explored. These all limit the further improvement in tracking performance. To address these issues, we propose a multi-scale cascaded single-object tracking framework based on spatio-temporal information fusion, which more comprehensively integrates the temporal and spatial information of targets and more deeply interacts the information in the search area and templates. In particular, to establish extensive spatio-temporal feature correlations, the spatio-temporal information fusion network is introduced, which uses a Transformer-based encoder-decoder structure to cross-fuse the global non-linear temporal and spatial information of the target search area with static and dynamic templates, effectively performing fusion-based propagation. To focus on rich spatial semantic information, we design a multi-scale feature extraction network, and a feature cascade aggregation module based on the encoder-decoder structure, which can effectively carry out interaction-based propagation. Finally, a bounding box prediction head and an IoU score head are used to predict the exact location of the target and update the dynamic templates respectively. Extensive experiments demonstrate that our method attains better tracking performance than the baseline method. Meanwhile, the proposed method also obtains comparable results with other state-of-the-art trackers on six challenging benchmarks, including GOT-10k, TrackingNet, LaSOT, UAV123, NFS30, and OTB100, while running at real-time speed.

In the field of autonomous driving, a novel approach is proposed for detecting and recognizing surrounding vehicles targets. The proposed method combines multiple attention mechanisms within a 3D object detection algorithm, drawing inspiration from the PointPillars network. Firstly, an additional local attention mechanism layer is incorporated into the point cloud pillar feature extraction module. This augmentation allows for a more comprehensive fusion of surrounding point cloud feature during the point cloud feature extraction process. Secondly, a channel attention mechanism module is introduced to the 2D image backbone network. The resulting features are subsequently inputted into the SSD algorithm for object detection. The improved algorithm's effectiveness has been demonstrated through an experimental evaluation on the popular KITTI dataset. The results show a significant improvement in average mean accuracy compared to the original algorithm, indicating that the attention mechanisms successfully enhance the detection and recognition performance of surrounding vehicles in autonomous driving scenarios.