MOJITO framework advances autonomous driving with unified sensor-to-action learning

Researchers have introduced MOJITO, a novel framework for end-to-end autonomous driving that utilizes modal joint learning. This approach bypasses the traditional two-stage pipeline, allowing the planning module to directly access and process multi-modal sensor data like images and LiDAR. MOJITO has demonstrated state-of-the-art performance on the NAVSIM v1 and NAVSIM v2 datasets, achieving 88.9 PDMS and 88.4 EPDMS respectively. The framework also shows promise in scalability, instruction following, and generating diverse trajectories. AI

IMPACT This unified framework could lead to more robust and adaptable autonomous driving systems by enabling direct use of multi-modal sensor data in planning.

RANK_REASON Research paper detailing a new framework for autonomous driving. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• arXiv
• Hugging Face
• MOJITO
• NAVSIM v1
• NAVSIM v2

• paper
• product
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →