Robotera's VPP2 Achieves Top Ranking on RoboDojo Simulation Benchmark for Robot Actions
- Published
- Source
- HumanoidsDaily
Robotera's Video Prediction Policy 2 (VPP2) has secured the top position on RoboDojo's simulation leaderboard, announced on October 9. This model effectively links scene change predictions with corresponding robot actions, outperforming PhysicalRSI in both average score and task success rate. VPP2 achieved a simulation score of 39.26, compared to PhysicalRSI's 36.27, while their success rates were 32.26% and 31.38%, respectively.
The significance of this achievement lies in VPP2's ability to enhance robot manipulation capabilities, as evaluated by the RoboDojo benchmark, which assesses generalization, precision, and long-horizon tasks. Despite its leadership, VPP2's performance on open-ended tasks was lower at 4.08%, highlighting the challenges even leading systems face. This nuanced performance data provides valuable insights into the capabilities of current robotic models.
Looking ahead, Robotera's ongoing research aims to refine VPP2's capabilities further, with reported success rates of 58.5% on the ALOHA platform and 45.0% on LIBERO-Pro. The release of Stage-1 and Stage-2 video-model weights on October 10, along with comprehensive training and evaluation guides, positions VPP2 as a significant resource for researchers in the field of robotic manipulation.
Original report
Robotera’s VPP2 Tops RoboDojo Simulation Benchmark by Linking Video Prediction to Robot Actions
HumanoidsDaily · [email protected] (Humanoids Daily Staff)This briefing is an independently written summary based on publicly available reporting and is provided for industry information and news discovery. The original report and source publication are credited and linked where applicable. RobotToday does not claim ownership of third-party source material.
Rights concerns? Contact [email protected] with the relevant URL and details. We will review the matter and take appropriate action where warranted.
More related news
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
Related Suppliers
Robotera 北京星动纪元
Beijing embodied AI company founded 2023 and backed by Tsinghua University; builds the L7 and STAR1 humanoids, Q5 wheeled robot and XHAND1 dexterous hand.
Yuejia Lingdong (Shenzhen) Technology Co., Ltd. 越甲灵动（深圳）科技有限公司
China
Shenzhen quadruped-robot maker that self-develops its own motors, reducers and controllers for grid inspection and security patrol robots.
Simate
China
Physical AI startup founded in 2026 whose Simate-beta robot model topped the RoboDojo leaderboard; also builds the Sinfra training platform.
Knowin 深圳诺因智能有限公司
China
Shenzhen home-robot startup founded 2025 building the foldable dual-arm KNOWIN-X1 and GLOW embodied model; raised a CNY 500M angel round in 2026.
Runway AI, Inc.
United States
US AI research company building general world models (GWM-1) for video generation and a robot-learning platform for policy inference, sim evaluation and data.
NC AI Co., Ltd.
Korea, Republic of (South)
NCSOFT's AI affiliate building the VARCO generative-AI suite and a World Foundation Model for robot intelligence and physical-AI synthetic data.
Generalist AI
United States
Frontier AI research company building embodied foundation models for general-purpose robots, with GEN-1 achieving 99% task success rates.
Dyna Robotics
United States
US startup building embodied AI foundation models for dual-arm commercial robots, achieving 99%+ success in 24/7 deployments in restaurants and laundromats.
Mbodi AI
United States
New York City embodied-AI startup, Y Combinator Spring 2025 batch; lets industrial robots learn manipulation tasks from spoken instruction and demonstration.
Kyber Labs
United States
Brooklyn, New York robotics startup building AI-controlled, mechanically compliant dexterous robot hands for assembly and manipulation tasks.
AMI Labs
France
AMI Labs is a Paris AI startup founded by Yann LeCun in 2025, building JEPA-based world models trained on real-world sensor data, not language.
BAAI 北京智源人工智能研究院
China
Beijing-based non-profit AI research institute developing large foundation models and open-source embodied-AI systems for robotics.
Robotics needs a service framework.
RSF defines a common language for robot service capability, lifecycle operations, certification pathways, and service-provider networks.
Join the RobotToday community on LinkedIn
Daily robotics news, in-depth analysis, conference highlights, and discussions with professionals worldwide.
