Prediction and Behaviors for Driver Assistance and Socially Cooperative Autonomous Driving
archive: archived pipeline: cataloged verified
Get this paper ↗ (full text — opens at the source; we link to it, we don't host it)
Summary
This paper addresses the challenge of generating socially cooperative and safe behaviors for autonomous vehicles in complex traffic scenarios, specifically focusing on unsupervised roundabouts and lane changes. The motivation stems from the need for autonomous systems to negotiate with human-driven vehicles effectively, ensuring fluidity and safety in mixed-traffic environments where driver intentions are uncertain. The authors aim to improve prediction and behavior planning modules to handle the high variance in road geometry and perception noise inherent in these scenarios. The study employs two distinct methodological approaches. For roundabout navigation, the authors formulate the problem as a Partially Observable Markov Decision Process (POMDP) to account for unknown driver intentions and perception uncertainties. They utilize the AR-DESPOT online, anytime planning algorithm, which represents the belief state using random particles to sample obstacle vehicles' motion intentions. This approach integrates prediction and planning into a single framework, optimizing for collision avoidance and adherence to target velocity. For lane changes, the authors propose a learning-based approach using non-parametric regression in Reproducing Kernel Hilbert Space (RKHS). This method treats the behavior generator as a continuous function that maps the past trajectories of the ego vehicle and five surrounding vehicles to the optimal start and end points of a lane-change maneuver. In the roundabout experiments, simulations were conducted using Python and MATLAB environments with varying numbers of participating vehicles. Results from 100 continuous runs demonstrated that the POMDP-based planner successfully guided the ego vehicle safely, preferring deceleration over lane changes due to higher penalties associated with the latter. The system scaled well with vehicle density, though behavior aggressiveness was highly dependent on the reward function tuning. For the lane-change study, the authors trained and tested their RKHS model on 543 scenarios extracted from the NGSIM dataset (I-80 and US-101 highways). The model achieved real-time performance with an average update time of 0.09 seconds. Experimental results indicated that the proposed inverse multiquadric kernel outperformed standard Laplacian and Gaussian RBF kernels in predicting feasible, human-like lane-change start and end points. The significance of this work lies in its contribution to socially cooperative autonomous driving. By integrating prediction and planning via POMDPs for roundabouts and using data-driven RKHS regression for lane changes, the authors provide methods that enhance the predictability and safety of autonomous vehicle interactions. The findings suggest that these approaches can generate natural, human-like behaviors that facilitate smooth traffic flow in mixed-autonomy settings, addressing critical gaps in current behavior planning systems that often struggle with unsupervised intersections and complex multi-agent negotiations.
Key finding
The AR-DESPOT POMDP algorithm enables safe autonomous navigation in roundabouts by balancing collision avoidance and target velocity adherence, while the RKHS non-parametric regression method accurately predicts lane-change start and end points using surrounding vehicle trajectory data.
Methodology
mixed_methods
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via bulk_ingest_rosap on 2026-05-23 (6 acquisition events logged).
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | rosap | — | — | 2 | 2026-05-23 |
| archive | success | — | — | — | 1 | 2026-05-23 |
| extract | success | cached | — | — | 2 | 2026-06-10 |
| clean | success | — | — | — | 1 | 2026-06-01 |
| chunk | success | — | — | — | 1 | 2026-06-01 |
| embed | success | — | — | — | 1 | 2026-06-02 |
| enrich | success | — | — | — | 1 | 2026-05-23 |
| promote | success | — | — | — | 1 | 2026-05-23 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 3 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 19 | 2026-06-11 |
| verify | success | — | — | — | 2 | 2026-06-10 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Methodological Resource: tool software
- Theoretical Contribution: computational model