Awesome Robots Digest - Issue #7 - October 11, 2025
Bob Jiang
October 11, 2025
TL;DR; 📋
Strategic Moves & Scale Bets: SoftBank acquires ABB Robotics for $5.4B • DeepMind launches Gemini Robotics 1.5 reasoning models • Serve Robotics expands with DoorDash partnership • NVIDIA releases Neural Robot Dynamics research
Introduction 🚀
Last week was dominated by strategic realignment and scale bets: SoftBank's big move into robotics, delivery bots expanding reach, and advances in robot "thinking" models from DeepMind. On the research side, NVIDIA's team released Neural Robot Dynamics, a promising work for more accurate dynamics modeling. Expect turbulence as these bets ripple downstream.
Top News & Breakthroughs 📰
🏢 Company News & Acquisitions
- SoftBank Acquires ABB's Robotics Division for $5.4B - SoftBank struck a deal to acquire ABB's robotics business, aiming to blend ABB's industrial strength with SoftBank's AI ambitions. The acquisition signals a move from SoftBank into industrial-scale robotics infrastructure, not just startups.
🚀 Product Launches & Partnerships
-
Serve Robotics Partners with DoorDash - Previously tied to Uber Eats, Serve Robotics expanded its deployment by partnering with DoorDash. This means more delivery bots hitting sidewalks in wider geographies.
-
DeepMind Launches Gemini Robotics 1.5 / ER 1.5 Models - DeepMind published two new models—Gemini Robotics 1.5 and Gemini Robotics-ER 1.5—built to improve robot reasoning and control in complex tasks. In demos, robots sorted fruit, explained their decision process, and generalized across embodiments more robustly.
🌐 Industry Developments
- "The Robot in Your Kitchen" — Figure 03 Hype & Reality - TIME ran a feature on Figure AI's ambition for Figure 03, a humanoid targeted at homes. Upgrades include tactile finger pads, enhanced perception, memory, and voice interaction. But the article also balances hype with caution: safety, cost, and social impact remain key unknowns.
Research Spotlight 🔬
📄 Research Papers
- Neural Robot Dynamics by NVIDIA Seattle Robotics Lab - A new paper presents Neural Robot Dynamics, a learned dynamics model designed to better predict robot motion and forces by blending data-driven and physical priors. This can reduce model error in simulation and improve real-world control fidelity.
🔧 Recent arXiv Submissions
-
U2UData-2 - A long-horizon autonomous flight dataset for UAV swarms, enabling better training and evaluation of multi-agent aerial systems.
-
Correspondence-Free Sim-to-Real for Deformable Surfaces - New methods for controlling soft objects without explicit planning models, advancing the field of deformable object manipulation.
Event Horizon 📅
🗓️ Recent Events
- DSEI 2025 - London - At DSEI 2025 in London, defense and robotics overlap heavily: bomb disposal robots, UAVs, autonomous ground vehicles, and robotics security systems were showcased.
Tool/Resource of the Week 🛠️
🎯 Featured Resource: Neural Robot Dynamics
A methodological breakthrough for more accurate robot dynamics modeling, bridging the sim-to-real gap.
Key Features:
- Blends data-driven learning with physical priors for improved accuracy
- Reduces model error in simulation environments
- Enhances real-world control fidelity
Why It's Useful: A better dynamics model can power more accurate simulation, policy transfer, error correction, and closed-loop control. For labs pushing into real robot deployments, this reduces the "sim-vs-real gap" that has long plagued robotics research.
Use Cases:
- Simulation-based training with better transfer to real robots
- Improved closed-loop control through accurate force prediction
- Reduced calibration time for new robot platforms
Community Corner 👥
💬 Trending Discussions
-
SoftBank/ABB Acquisition - The acquisition is generating speculation over consolidation: Will we see more conglomerates absorbing robotics divisions? The community is divided on whether this signals maturation or concentration of power.
-
Delivery Bot Expansion - Serve's move to partner with DoorDash triggered chatter: delivery bots may finally cross threshold from pilots to scale. Urban planning and regulatory discussions are heating up.
🌟 Hot Topics
-
Gemini Robotics Models - The new versions sparked discussion in RL/robotics forums: how far can vision + language + dynamics go before needing specialized modules? Researchers are debating the balance between general and specialized capabilities.
-
Figure 03 Profile - TIME's article stirred debate: some see "robot in the home" as plausible in the next decade; others warn of socioeconomic and safety ramifications. The gap between vision and reality remains a contentious topic.
Trends to Watch 📊
-
Big tech & industrial tie-ups - SoftBank moving heavy into robotics infrastructure suggests the next wave will be about ownership and control of the platform stack (hardware + AI).
-
Robotic "reasoning" models maturing - Gemini 1.5 / ER upgrades push the frontier of what robots can think, not just react.
-
Edge dynamics models - Learned models like Neural Robot Dynamics give robots the capacity to predict and adapt better in real environments.
-
Scaling delivery robots - Serve's shift to DoorDash shows delivery bots may finally gain commercial leverage beyond niche pilots.
-
Humanoids as lifestyle speculation - Figure 03 is less a product today than a vision of robots in homes. But vision shapes investment, development, and expectations.
Conclusion 🎯
Issue #7 highlights strategic moves more than flashy demos. The SoftBank acquisition, Gemini model update, and delivery robot expansion suggest this week will be looked back on as a structural inflection point. Research is keeping pace—Neural Robot Dynamics being a key example—but the balance between hype and durable capability will be tested in the coming months.
What matters most to you in these developments? Are you tracking humanoids, logistics automation, or control theory advances? We'd love to hear your perspective as these trends unfold. Stay tuned for next week's issue, where we'll continue tracking the ripple effects of these strategic shifts.
📧 Stay Connected
- Subscribe: Newsletter signup link
- Follow us: Twitter (X) links
- Website: Official Website
This digest is curated by the Awesome Robots team. Have a story to share? Contact us