2026· IEEE Transactions on Automation Science and Engineering· Vol 23, pp. 13123-13135· 0 citations· 53 references
Abstract
Accurate, real-time motion forecasting is crucial for safe operation of autonomous vehicles. State-space models, such as Mamba, can model long-range dependencies with linear computational complexity, reducing sequence modeling overhead. However, Mamba-based motion forecasting methods still rely on Transformer-style decoding frameworks, which have high computational complexity and struggle to meet real-time requirements in complex traffic scenarios. To address this, this paper proposes a multimodal motion forecasting method IPDM, based on the intent-parametrized dynamics Mamba model. First, we design the decoupled intent-state parameterization module, which decomposes multimodal queries into intent and state queries, constructs specific state-space model parameters, and generates multimodal trajectories with linear computational complexity, thereby reducing computational complexity of the decoding stage. Furthermore, addressing the challenge that single-time-scale state-space models within the Mamba decoding framework struggle to balance long-term memory with instantaneous maneuvering responses, we propose a dual-scale controlled scan mechanism. This mechanism constructs short-term and long-term flows and employs intent-aware gated adaptive fusion to enhance forecasting performance while maintaining linear decoding properties. Experimental results on the Argoverse 1 and Argoverse 2 datasets demonstrate that IPDM achieves state-of-the-art performance in inference speed and parameter efficiency, outperforming most existing methods while reducing the number of parameters by over 40%. Note to Practitioners—In autonomous driving systems, vehicles must determine the likely trajectories of surrounding road users over the next few seconds within a very short timeframe to support path planning, obstacle avoidance, and safety decisions. However, in practical deployment, motion forecasting modules must not only ensure accuracy but also meet the strict constraints of onboard edge hardware regarding computational resources and inference latency. While many existing multimodal prediction methods offer good performance, their decoding processes typically involve high computational overhead, making it difficult to balance accuracy and real-time performance in complex traffic scenarios. This work is motivated by the need to address this engineering challenge. The method proposed in this paper separates the modeling of “how the vehicle might want to move” from “how the trajectory evolves over time.” By doing so, it reduces the computational burden of the decoding stage while maintaining multimodal prediction capabilities, thereby enhancing the method’s potential for deployment on resource-constrained platforms. Experimental results demonstrate that the method maintains strong prediction performance while reducing both model parameters and inference overhead. It should be noted that the method has primarily been validated on standard public datasets; for real-world deployment, further consideration must be given to the impacts of perception noise, map errors, and extreme traffic conditions. Beyond autonomous driving, this approach is expected to be applicable to delivery robots, roadside intelligent devices, and other embedded automation systems requiring rapid motion forecasting.
An efficient Mamba-based feature extraction framework for jointly encoding vehicle trajectories and map information is proposed and achieves superior performance in terms of minADE, minFDE, and minMR, while maintaining high computational efficiency.
J. Li, L. Wang, J. Pei· Revista Internacional de Mét...· 0 citations
Trajectory prediction is essential for many robotic applications, yet most existing models rely on fixed-length observations and struggle with temporally irregular inputs. In real-world settings, prediction difficulty further increases when agents exhibit strong maneuverability, as their future motions depend on distinct short-term and long-term temporal cues. A Maneuver-aware Temporal Fusion framework is proposed to separate short-term dynamics from long-term intentions and fuse them through a motion-complexity-guided attention mechanism. The framework first extracts temporal features at different scales, and then adaptively balances them according to the maneuver patterns of each agent. To support incomplete or short observations, a self-distillation strategy is introduced to reconstruct missing motion segments, enabling consistent prediction without relying on explicit teacher-student models. Furthermore, a Mamba-Transformer hybrid backbone is employed to enhance computational efficiency and improve generalization under arbitrary observation lengths. Experiments on the ETH/UCY and SDD datasets show that MaTF consistently outperforms existing methods, particularly in scenarios with irregular or shortened observations.
Shuobo Wang, Wenyuan Qin, Yongzhao Hua et al.· IEEE Robotics and Automation...· 0 citations
Visual localization is vital for autonomous systems but remains challenging under dynamic conditions. Transformers offer strong temporal modeling at quadratic cost, while CNNs are efficient yet limited in long-range dependencies. Existing methods also lack robustness to illumination, weather, and seasonal changes, constraining real-world applicability. To address this, this paper proposes AdapseqNet, a dual-branch architecture that integrates stabilized state-space modeling with differential temporal enhancement. First, a stabilized state-space formulation featuring Lyapunov-constrained parameterization and adaptive discretization is proposed, ensuring asymptotic stability and linear computational complexity for reliable processing of extended sequences. Second, a selective Mamba architecture is developed to combine temporal-state modeling with content-aware gating, enabling adaptive feature selection that emphasizes discriminative cues while suppressing redundancy. Third, a differential enhancement module is designed to extract motion-invariant representations through symmetric temporal differencing and LSTM-based refinement, enhancing resilience to appearance variations caused by lighting, weather, and seasonal changes. Beyond architectural design, multi-scale feature fusion and output distribution control are incorporated to optimize representation quality and ensure consistency for similarity-based retrieval. Extensive experiments on multiple benchmarks demonstrate that AdapseqNet achieves a better localization accuracy across diverse and challenging conditions. Note to Practitioners—Visual localization is crucial for autonomous robots but often fails under varying lighting, weather, or seasonal conditions. We propose a dual-path approach: one path captures long-term patterns using control-inspired stable modeling, while the other extracts motion cues that remain consistent despite appearance changes. This combination enables accurate place recognition even in extreme environments. Our system operates efficiently on standard hardware and was tested on an indoor robot, achieving centimeter-level accuracy. This approach can enhance existing navigation systems without requiring additional sensors. Future work will focus on real-time optimization for outdoor deployment.
Zhenyu Li, Tianyi Shang· IEEE Transactions on Automat...· 0 citations
A Stochastic Gating Decoder for multimodal latent variable sampling, adaptively fusing kinematics and data-driven paths to capture driver intention uncertainty while maintaining kinematic consistency is introduced.
Long-horizon planning is critical for safe autonomous driving in complex scenarios. Existing methods improve planning continuity with temporal memory, but such memory may become invalid and mislead decisions when the driving command changes. Thus, selectively leveraging useful history while suppressing command-inconsistent memory remains a key challenge. To address this issue, we propose MomADv2, a reliable state-space memory framework for long-horizon end-to-end autonomous driving. At its core, MomADv2 introduces a Selective State-Space Planning Memory Query Module, which filters historical planning queries based on temporal continuity and command consistency, selects planning modes relevant to the current command, and models the evolution of planning intentions through a selective state-space mechanism. To further alleviate local trajectory deviations and error accumulation in long-horizon planning, we design a Flow-Matching Trajectory Residual Refiner. It learns a continuous residual correction field from the refined planning output to the expert trajectory, enabling fine-grained trajectory refinement while preserving the stability of anchor-based planning. Extensive experiments on closed-loop NAVSIM and Bench2Drive, as well as open-loop nuScenes, demonstrate that MomADv2 improves long-horizon planning consistency and reduces the average collision rate by 15.6% over MomAD under 6-second planning.
Ziying Song, Shengkai Zhang, Lin Liu et al.· 0 citations
Long-term ship trajectory prediction is a fundamental capability for maritime safety and autonomous navigation. While recent Transformer-based architectures have improved forecasting horizons, they predominantly rely on historical kinematic states, treating vessel motion as an isolated system. In reality, maritime navigation is profoundly modulated by extrinsic factors like weather and constrained by static vessel characteristics. Existing multimodal approaches fundamentally model the joint distribution over states and contexts, treating environmental variables as peer features rather than encoding the directional physical dependence of vessel dynamics on environmental conditions. In this work, we propose the Conditional Informer, a novel encoder-decoder architecture that formulates trajectory prediction as a conditional generation task. We employ a dedicated Conditional Attention mechanism where the vessel state explicitly queries environmental contexts through cross-attention, encoding the physical prior that weather modulates - but is not generated by - vessel dynamics. Furthermore, to address the intermittency of real-world data, we introduce a Modality Masking training strategy to prevent catastrophic degradation during sensor fallback. Extensive experiments on AIS and ERA5 data demonstrate that our approach outperforms kinematic and concatenation-based baselines by 15.4% in prediction accuracy when context is available. Crucially, Modality Masking prevents shortcut learning, reducing fallback error by nearly an order of magnitude compared to unconstrained models.
Yuansheng Guan, C. Squires, Timothy Hu et al.· 0 citations