• Mechonomics
  • Posts
  • NEURA Robotics & HD Hyundai team up to bring robots to shipyards

NEURA Robotics & HD Hyundai team up to bring robots to shipyards

+ Galbot raises ¥1.1 billion to commercialise its G1 semi‐humanoid

Author

Maddy Wong
August 08, 2025

🧠 Weekly Brief

1. NEURA Robotics & HD Hyundai team up to bring robots to shipyards

German robotics start‑up NEURA Robotics signed a strategic partnership with HD Hyundai’s shipbuilding units to develop and test quadruped and humanoid robots in shipyards. The partners will validate cognitive robots under real‑world conditions to boost efficiency and safety in one of industry’s most challenging environments.

3 Key Takeaways:

  1. Shipyard trial program: NEURA, HD Hyundai Samho and HD Hyundai Robotics will field test quadruped and humanoid robots for tasks such as welding, inspection and material handling, aiming to increase productivity and safety.

  2. Cognitive robot versatility: NEURA founder David Reger said the partnership demonstrates how cognitive robots can handle demanding industrial environments.

  3. Korean commitment: Executives from HD Hyundai emphasised that specialised welding robots could improve quality and productivity in shipbuilding.

These trials signal a shift from flashy demos to practical industrial deployment, showing how robotics firms are teaming with legacy manufacturers to solve labour shortages and safety challenges in heavy industry.

2. Galbot raises ¥1.1 billion to commercialise its G1 semi‑humanoid

Beijing‑based Galaxy General Robot Co. (Galbot) closed a RMB 1.1 billion (≈US$153 million) funding round led by CATL and Puquan Capital. The funding brings total investment to about US$335 million and will accelerate mass deployment of its G1 semi‑humanoid mobile manipulator designed for inventory, delivery and packaging tasks.

3 Key Takeaways:

  1. Rapid commercialisation: The G1 robot is already operating in nearly 10 Beijing stores and can be deployed in a day; Galbot plans to expand to 100 stores by year‑end.

  2. Vision‑language‑action models: Galbot developed proprietary GraspVLA, GroceryVLA and TrackVLA models enabling zero‑shot grasping and scene adaptation for 5,000 different goods.

  3. Strategic investors: The round was led by battery giant CATL and Puquan Capital, with participation from China Development Bank and Qiming Venture Partners underlining interest in embodied intelligence from energy and state‑backed investors.

The raise underscores investors’ appetite for embodied AI platforms that can perform practical tasks today, offering an intermediate step between wheeled robots and full humanoids.

3. Amazon deploys its millionth robot and unveils DeepFleet AI model

Amazon announced that its one‑millionth robot was delivered to a fulfilment centre in Japan, cementing it as the world’s largest mobile‑robot operator. The company’s diverse fleet now includes Hercules units that lift 1,250 pounds and Proteus autonomous mobile robots; Amazon also unveiled DeepFleet, a generative AI model to coordinate robot movements.

3 Key Takeaways:

  1. Scale milestone: A million robots now assist 75 % of Amazon’s global deliveries, reflecting the company’s decade‑long investment in automation.

  2. New AI logistics brain: DeepFleet optimises robot travel paths, reducing congestion and improving efficiency by 10 %.

  3. Human‑robot balance: Despite employing 1.56 million people, Amazon’s CEO signalled workforce reductions as generative AI takes over more day‑to‑day tasks, highlighting the tension between automation and human labour.

Amazon’s milestone demonstrates how large‑scale fleets and AI scheduling models are redefining warehouse operations and foreshadowing further shifts in retail logistics.

🤖 Startup Spotlight

Galbot: Funding fuels semi‑humanoid revolution

  • Location: Beijing, China

  • What they do: Galbot develops semi‑humanoid robots that move on wheels and use arms to automate retail tasks such as inventory checks, replenishment and delivery. The G1 can handle 5,000 product types and deploy in new stores within a day.

  • Latest update: Raised RMB 1.1 billion (~US$153 million) in July 2025 led by CATL and Puquan Capital, bringing total funding to about US$335 million.

Why watch: The funding boost accelerates G1 deployments across China and highlights how semi‑humanoids—combining mobility with dexterous arms—offer a commercially viable bridge between wheeled robots and fully bipedal humanoids. Galbot’s proprietary vision‑language‑action models enable zero‑shot generalisation and fast adaptation, positioning the company at the forefront of embodied AI retail solutions.

📈 Investor Watch

Construction robotics: optimism surges but deployments lag

BuiltWorlds’ 2025 Equipment & Robotics Benchmarking Report found that contractors, owners and tech providers increasingly view construction robots as essential for creating more efficient and scalable job sites. Positive evaluations of robotics strategies jumped from 74 % in 2024 to over 95 % this year, while negative attitudes disappeared. However, the share of companies actually using robots dropped from 65 % to 46 %. BuiltWorlds notes that last year’s numbers were inflated by one‑off pilots; this year sees fewer pilots but more repeated use on projects, indicating selective but serious adoption.

Key takeaways:

  • Investor enthusiasm for construction‑automation startups is justified by soaring industry interest, but real‑world adoption remains in early stages.

  • Declining pilot activity suggests investors should favour companies focused on repeatable, scalable deployments rather than proofs of concept.

  • As repeated use cases emerge, the sector could transition from “future tech” to a standard jobsite tool, offering long‑term upside.

🧩 Pattern of the Week

The adoption paradox: High enthusiasm, cautious implementation

Across industries, robotics is enjoying record enthusiasm—shipbuilders partner with cognitive‑robotics firms, retailers raise nine‑figure funding rounds for semi‑humanoids and warehouses deploy million‑strong robot fleets. Yet BuiltWorlds reports a drop in construction‑robot deployments despite surging positive sentiment. This “adoption paradox” suggests that enterprises are excited about robotics’ potential but remain selective in implementation, focusing on repeatable use cases rather than pilots. For the robotics industry, converting enthusiasm into deployment will hinge on delivering reliable, scalable solutions and quantifiable ROI.

📚 Resource / Reading

Humanoids, AI chips and autonomy: IDTechEx maps the future of robots

Edge AI and Vision Alliance summarises key findings from IDTechEx’s report “Humanoid Robots 2025–2035”. The article explains how LiDAR and 3D vision give humanoid robots situational awareness, enabling them to assemble car parts, handle materials and perform inspections. It also envisions light‑duty robots performing routine health checkups in hospitals to ease staffing pressures. Soft‑gripper technology allows robots to handle delicate items like fruit, while AI and voice recognition enable them to understand spoken commands and collaborate with humans. Edge computing and specialised AI chips bring processing power closer to robots, reducing latency and supporting autonomy.

Why read: This resource provides a market‑research perspective on technological enablers (sensors, soft grippers, edge AI) that will drive humanoid and collaborative robots into industries, homes and healthcare over the coming decade. It complements this newsletter’s news items by highlighting underlying technology trends and market projections.

🛠 Builder’s Corner

Multi‑agent AI: Designing the brains of robot fleets

Field AI is pioneering Field Foundation Models to coordinate fleets of robots operating in unstructured, high‑risk environments. Its Multi‑agent AI Research Engineer role focuses on developing game‑theoretic and decentralised algorithms for multi‑robot coordination. The engineer designs scalable algorithms for mean‑field control, Nash equilibria and auction‑based task allocation while integrating these into real‑world hardware. Compensation ranges from $70 k–$200 k depending on experience.

What it teaches builders: The posting highlights how coordinating multiple robots is an emerging challenge requiring expertise in game theory, decentralised optimisation and reinforcement learning. It underscores the need to bridge theoretical research with practical deployment—an essential skillset as companies deploy fleets of robots across logistics, agriculture and defence.

💼 Jobs in Mechonomics (Most Recent Humanoid‑Robot Roles)

Multi‑agent AI Research Engineer – Field AI | Mission Viejo, CA / Boston, MA | $70 K–$200 K

Develop algorithms for multi‑robot fleet coordination at Field AI. You’ll design and implement game‑theoretic and decentralised control strategies, build predictive models for multi‑agent interaction dynamics and integrate auction‑based task allocation into real hardware. The role blends theory and deployment and offers hybrid or on‑site optionsjobs.lever.co.

Senior Robotics Manipulation Engineer – Dexterity | Redwood City, CA | $170 K–$225 K

Join Dexterity’s mechanical‑engineering team to build robot manipulation models that pick, move and pack goods. Responsibilities include training models for force control and servoing, converting heuristic controllers to learned algorithms and collaborating with data, simulation and operations teams. Strong Python/C++ skills and experience with reinforcement learning and motion planning are essentialjobs.lever.co.

Sr. System Integration Engineer – Reliable Robotics | Mountain View, CA | $162 K–$220 K

Work on automated aviation systems that combine mechanical, electrical and software components. You’ll own vehicle‑level projects from flight‑control hardware to hardware‑in‑the‑loop testing, manage integration schedules and prioritise aircraft functions. The job requires eight or more years of electromechanical integration experience and offers stock options and benefitsjobs.lever.co.

Avionics Hardware Engineer – Reliable Robotics | Mountain View, CA | $110 K–$168 K

Join the Avionics Hardware team to design and troubleshoot circuit‑board‑level electronics for autonomous aircraft. Tasks include developing compute, actuation and sensing hardware, conducting benchtop and environmental testing, and collaborating across electrical, mechanical and software teams. Experience with mixed‑signal circuit design and ECAD tools is preferredjobs.lever.co.

Data Collection Operator – Tesla Optimus | Palo Alto, CA | $25.25–$48 per hour

Help train Tesla’s Optimus humanoid by wearing motion‑capture suits and performing movements that robots can mimic. Operators move and pose with the robot under controlled conditions to collect high‑quality data for motion‑planning algorithms. The role pays $25.25–$48 per hour and provides a pathway into cutting‑edge embodied AI researchfortune.com.

Source (via the article on Tesla hiring Optimus data‑collection operators)