CooperDrive: Enhancing Driving Decisions Through Cooperative Perception

Deyuan Qu; Qi Chen; Takayuki Shimizu; Onur Altintas

cs.ROApr 15, 2026

CooperDrive: Enhancing Driving Decisions Through Cooperative Perception

Authors: Deyuan Qu, Qi Chen, Takayuki Shimizu, Onur Altintas

Organizations: Toyota InfoTech Labs, Mountain View, CA, USA

Abstract

Autonomous vehicles equipped with robust onboard perception, localization, and planning still face limitations in occlusion and non-line-of-sight (NLOS) scenarios, where delayed reactions can increase collision risk. We propose CooperDrive, a cooperative perception framework that augments situational awareness and enables earlier, safer driving decisions. CooperDrive offers two key advantages: (i) each vehicle retains its native perception, localization, and planning stack, and (ii) a lightweight object-level sharing and fusion strategy bridges perception and planning. Specifically, CooperDrive reuses detector Bird's-Eye View (BEV) features to estimate accurate vehicle poses without additional heavy encoders, thereby reconstructing BEV representations and feeding the planner with low latency. On the planning side, CooperDrive leverages the expanded object set to anticipate potential conflicts earlier and adjust speed and trajectory proactively, thereby transforming reactive behaviors into predictive and safer driving decisions. Real-world closed-loop tests at occlusion-heavy NLOS intersections demonstrate that CooperDrive increases reaction lead time, minimum time-to-collision (TTC), and stopping margin, while requiring only 90 kbps bandwidth and maintaining an average end-to-end latency of 89 ms.

Figures & tables

Fig. 1 : Autonomous vehicles inevitably face sensing limitations due to occlusions and blind spots. Cooperative perception addresses this challenge by allowing vehicles to share complementary observations, enabling earlier awareness of potential hazards and supporting safer driving decisions.

Fig. 2 : Reconstructed Bird’s-Eye View (BEV) representation generated through cooperative perception, where each vehicle contributes precise localization and detection outputs from our proposed Multi-task BEV Perception Network to enhance situational awareness and support safer path planning.

Fig. 3 : Comparison of a left-turn intersection scenario (real-world) under ego-only perception and CooperDrive over time (t0 to t10). Top: With ego-only perception, the oncoming vehicle from the left remains occluded and undetected until t9 when it enters the collision zone, leading to a potential collision risk. Bottom: With cooperative perception, the sender detects the dynamic vehicle from t0 and continuously shares it with the ego vehicle. This enables the ego to maintain awareness throughout the sequence and plan an early stop before the collision zone, ensuring safety.

Error Level	Translation error (m) $\downarrow$			Heading error (deg) $\downarrow$
Error Level	GNSS	ICP	NDT	GNSS	ICP	NDT
1	2.26	1.57	0.52	4.47	1.26	0.74
2	3.41	1.96	0.71	6.68	1.53	0.96
3	4.51	2.76	1.32	8.97	4.53	1.90
4	5.61	4.03	2.12	11.26	8.24	2.82

TABLE I : Localization evaluation under noisy GNSS initialization on the nuScenes dataset. We compare the noisy GNSS prior with ICP and NDT scan-to-map pose refinement using the learned current keypoints and the global geometry map.

Model	Backbone	mAP $\uparrow$	NDS $\uparrow$
SSN [ 23 ]	SECFPN	35.17	49.76
PointPillars [ 24 ]	SECFPN	35.19	50.27
CenterPoint [ 25 ]	SECFPN	56.11	64.61
Ours	SECFPN	56.27	64.78

TABLE II : LiDAR 3D object detection on nuScenes dataset.

Method	$\mathrm{TTC}_{\min}$ (s) $\uparrow$	$\mathrm{DRAC}$ (m/s 2 ) $\downarrow$	$\mathrm{DCZ}$ (m) $\uparrow$	VR (%) $\downarrow$
Ego-only	$2.05$	$0.312$	$2.01$	$18$
CooperDrive	$6.62$	$0.060$	$4.12$	$2$
Improvement	+4.57	-0.252	+2.11	-16

TABLE III : Overall planning metrics comparison across all test scenarios: Ego-only vs. CooperDrive.

Fig. 4 : Comparison at a T-intersection right turn: Ego-only vs. CooperDrive.

Fig. 5 : CooperDrive at a T-intersection right turn with a sender emerging from the ego’s blind spot.

Fig. 6 : Real-world deployment analysis of communication efficiency and end-to-end latency.

Cooperative Perception

Cooperative perception lets connected vehicles, robots, and aerial agents combine sensor observations to understand scenes beyond a single viewpoint.

See all 111 papers

CoCam4D: Geometry-Aware Cooperative 4D Perception for Camera-Only Autonomous Driving
Oct 8, 2026Soham Pahari, Sudip Das, Arindam Das +1Autonomous Driving Perception Cooperative Perception
University of Petroleum and Energy Studies, Dehradun · Indian Statistical Institute, Kolkata · anSWer India, Valeo Brain
DH-VLM: Dual-Horizon Cooperative Latent Reasoning for Autonomous Driving
Aug 10, 2026Ziyi Song, Chen Xia, Hang Yu +2VLMs for Autonomous Driving Autonomous Driving
Department of Electronic Engineering, Tsinghua University · State Key Laboratory of Intelligent Green Vehicle and Mobility, Tsinghua University
UECP: Uncertainty-Enhanced Collaborative Perception
Jun 22, 2026Kang Yang, Tianci Bu, Peng Wang +3Uncertainty Quantification Autonomous Driving Perception
Renmin University of China, Beijing, China · College of Systems Engineering, National University of Defense Technology, Changsha, China · Intelligent Shipping Center, China Waterborne Transport Research Institute, Beijing, China

Oct 8, 2026J/K move · Enter open · S save

CoCam4D: Geometry-Aware Cooperative 4D Perception for Camera-Only Autonomous Driving

Soham Pahari, Sudip Das, Arindam Das, Ujjwal Bhattacharya

University of Petroleum and Energy Studies, Dehradun · Indian Statistical Institute, Kolkata · anSWer India, Valeo Brain

Autonomous Driving Perception Cooperative Perception 3D Scene Understanding 4D Scene Understanding Communication-Efficient Collaborative Perception

Open paper AskPDF arXiv

Autonomous vehicles often suffer from limited perception due to occlusions, blind spots, limited sensor range, and the complex nature of surrounding environments. Multi-agent collaborative perception (CP) addresses these challenges by allowing vehicles to share sensory information and reconstruct the scene cooperatively. However, camera-only perception remains fundamentally limited by the uncertainty of distance-dependent monocular depth estimation. We propose CoCam4D, a Bayesian framework for collaborative perception that explicitly models geometric uncertainty. It uses a VGGT-based feedforward network to generate 3D Gaussian scene representations with associated uncertainty estimates, enabling multiple vehicles or agents to efficiently combine their observations. By sharing compact Gaussian primitives, reliable observations from one agent can reduce the depth uncertainty of another without requiring LiDAR sensors. To support real-world deployment, we introduce Dynamic Object Primitives (DOPs), a compact 35-byte representation designed for efficient C-V2X communication. Extensive experiments show that our proposed method consistently outperforms recent vision-only methods, achieving improvements of 11.48% on OPV2V+ and 10.62% on DAIR-V2X-C, demonstrating the potential of geometrically grounded collaborative perception for LiDAR-free autonomous driving.

Abstract

Cooperative Perception

Explore similar work

CoCam4D: Geometry-Aware Cooperative 4D Perception for Camera-Only Autonomous Driving

DH-VLM: Dual-Horizon Cooperative Latent Reasoning for Autonomous Driving

UECP: Uncertainty-Enhanced Collaborative Perception