Autonomous driving technology has garnered significant attention due to its potential to revolutionize transportation through advanced robotic systems. Despite optimistic projections for commercial deployment, the development of sophisticated autonomous driving systems remains largely experimental, with the effectiveness of neurorobotics-based decision-making and planning algorithms being crucial for success. This paper delivers a comprehensive review of decision-making and planning algorithms in autonomous driving, covering both knowledge-driven and data-driven approaches. For knowledge-driven methods, this paper explores independent decision-making systems, including rule based, state transition based, game-theory based methods and independent planing systems including search based, sampling based, and optimization based methods. For data-driven methods, it provides a detailed analysis of machine learning paradigms such as imitation learning, reinforcement learning, and inverse reinforcement learning. Furthermore, the paper discusses hybrid models that amalgamate the strengths of both data-driven and knowledge-driven approaches, offering insights into their implementation and challenges. By evaluating experimental platforms, this paper guides the selection of appropriate testing and validation strategies. Through comparative analysis, this paper elucidates the advantages and disadvantages of each method, facilitating the design of more robust autonomous driving systems. Finally, this paper addresses current challenges and offers a perspective on future developments in this rapidly evolving field.

Lane detection has been one of the most important functions in the autonomous driving perception module. Most of the current research require complex post-processing and curve fitting processes before they can be used by subsequent regulation modules. In this paper, we propose the LLFormer algorithm combining CNN and Transformer structure, which is the first attempt to perform end-to-end lane detection based on laser point cloud and output its cubic polynomial coefficients. In addition, this paper modifies the structure of the conventional transformer and proposes the Generating Lane Query (GLQ) module. The output of encoder is plugged into GLQ for initialization of lane query in decoder, preserving the uniqueness of each frame of point cloud data. We test the performance of the proposed algorithm in the public dataset K-Lane, and the results show that the accuracy of the proposed LLFormer is close to the existing SOTA algorithm. The number of model parameters of LLFormer is only 9.01M, and the amount of operations is only 0.19GFLOPs, which are 1/26 and 1/2937 of the existing SOTA algorithm, respectively. The frequency of inference calculation is 35.9FPS, which can fully meet the real-time requirements for industrial deployment.

Detection based on LiDAR and camera fusion is increasingly popular for researchers in the autonomous driving domain. Compared to the camera-only and LiDAR-only methods, the fusion-based methods indeed improve the detection accuracy on public-available datasets. However, due to the complexity of the projection or fusion mechanism, few of these methods can run in real time even on an advanced desktop GPU. Thus, in this paper, we propose a new fusion detection model G-Fusion with a light and fast image view-transform module. According to our receptive field analysis of image feature maps, we directly project image features to only one voxel layer located on the ground, then fuse the LiDAR and image features by concatenation and convolution. With this delicately designed module, G-Fusion greatly boosts the state-of-the-art speed performance on the nuScenes dataset, achieving a good balance with the competitive detection scores. Meanwhile, since the precision of sensor extrinsic parameters is important for most fusion-based methods, we also deeply dig into our model’s calibration error tolerance ability and discover the failure noise condition.

The letter introduces a novel Visual Place Recognition (VPR) method called Lightweight Pooling-centric Saliency-aware VPR (LPS-VPR), a high-performance VPR method capable of exploiting saliency information without computational burden. The key contribution of the method is a pooling-based saliency encoder (PSE) that efficiently extracts and integrates local context saliency cues into the image embedding, avoiding using computationally intensive operations such as attention for saliency cues extraction. The method employs a multi-level feature pyramid fusion with a cascaded backtracking structure, merging multi-scale embedding information from different levels of the backbone network with negligible computational demand. The LPS-VPR method demonstrates superior performance on mainstream VPR benchmarks, showcasing state-of-the-art (SoTA) runtime efficiency and detailed ablation studies highlight the effectiveness of each LPS-VPR component. Extended visual demonstrations further underscore the effectiveness and superiority of LPS-VPR, making it a promising solution for real-world large-scale VPR applications.

Visual Place Recognition (VPR) is crucial for various computer vision and robotics applications. Traditional VPR techniques relying on handcrafted features, have been enhanced by using Convolutional Neural Networks (CNNs). Recently, MixVPR has set new benchmarks in VPR by using advanced feature aggregation techniques. However, MixVPR's full-image feature mixing approach can lead to the ignoring of critical local detail information and regional saliency information in large-scale images. To overcome this, we propose MixVPR++, which integrates an Adaptive Gabor Texture Fuser with a Learnable Gabor Filter for enriching semantic context with texture details information and a Hierarchical-Region Feature-Mixer for better spatial hierarchy capture regional saliency information, thereby enhancing robustness and accuracy. Extensive experiments demonstrate that MixVPR++ outperforms state-of-the-art methods across most challenging benchmarks. Despite its impressive performance, MixVPR++ shows limitations in handling severe viewpoint changes, indicating an area for future improvement.

Bird’s Eye View (BEV) perception requires swift and accurate representation of real-world 3D information, and BEV-based 3D object detection and semantic segmentation are foundational for downstream autonomous tasks such as prediction and planning. Current BEV perception methods are divided into Light Detection and Ranging(LiDAR)-based and camera-based methods depending on the input data modalities. LiDAR-based methods offer precise 3D information, but are limited by high sensor costs. Conversely, camera-based methods are economical and widely researched. However, 2D image features transform into BEV space introducing feature distortion due to inaccurate 3D information, particularly in depth estimation which results in noisy projections. Denoising Diffusion Probabilistic Models (DDPM) are used to mitigate noise in BEV features. However, DDPM increases computational costs, relies on noisy images BEV features, and limits effectiveness. To address these issues, we propose a lightweight BEV perception framework (DMPG-BEV) to enhance the performance of camera-based methods. This framework employs Denoising Diffusion Implicit Models (DDIM) to generate point cloud features with 3D information, while reducing computational demands by fewer denoising steps. We also introduce a lightweight feature fusion module which integrates point cloud BEV features with image BEV features to incorporate 3D information and reduce noise. The refined BEV features are also well-suited for multi-task networks involving detection and segmentation tasks. Additionally, a lightweight view transformation module is included to further reduce computational costs. Extensive experiments on the nuScenes benchmark demonstrate better performance of our proposed model with reduced computation, highlighting its value in BEV perception.