Mistral AI unveils first robotics model as Europe's AI leader pushes into physical automation

Mistral launches first robotics model in physical AI push

July 8 (Reuters) - Paris-based Mistral AI on Wednesday unveiled its first robotics model as Europe's leading AI company pushes into factories, warehouses and industrial automation. The launch follows Mistral's acquisition ⁠of Austria's Emmi AI in May and comes months after Paris-based startup Genesis AI unveiled a broader robotics model with navigation and manipulation capabilities. Reporting by Leo Marchandon in Gdansk; Editing by Matt Scuffham Our Standards: The Thomson Reuters Trust Principles., opens new tab

Mistral expands physical AI offering with first robotics launch

The organisation also announced it is actively expanding the robotics team and is looking to recruit talented research scientists and engineers. France's Mistral AI has announced the launch of a new robotics navigation model, as the company further expands in the physical AI space, following deals with a number of key players in Europe's industrial and manufacturing sector, such as Airbus SE and BMW. The new 8B model, Robostral Navigate, allows robots to autonomously move around in complex environments via a single RGB camera and basic language prompts. Combining pointing-based navigation with continuous learning elements, the hardware is also agnostic meaning it can be deployed across any robotics fleet. Mistral claims that the model, prompted by a single instruction, can complete the entire task on its own, moving through a live space full of people and obstacles it was never shown, adapting to any setting. Spaces in which it can be used includes offices, residential and commercial buildings and outdoor settings. In a post announcing the launch, Mistral said, "We leverage our knowledge of post-training LLMs at scale, using online reinforcement learning, to boost the performance of Robostral Navigate. After the supervised training stage, we further improve the model's performance using CISPO, an online reinforcement learning algorithm. "This enables the model to learn from trial and error, recover from failures, and acquire exploratory behaviours, effectively mitigating the distribution shift issue of vanilla behaviour cloning. This alone improved the success rate by 3.2pc. We are not seeing any plateauing, so we are confident that more training and more experiments will continue to push this number up." A leader in Europe's AI space, Mistral is also positioned as a key rival for US counterparts, such as Anthropic and OpenAI. In March of this year, the company raised $830m in its first debt financing, with the intention of funding a new data centre near Paris. It was announced that the deal, which was supported by a consortium of seven global banks, would pay for Nvidia Grace Blackwell infrastructure with 13,800 Nvidia GB300 GPUs at the "cutting-edge" centre, bringing powered capacity to 44MW. Previously commenting, Arthur Mensch, the CEO of Mistral AI, said, "Scaling our infrastructure in Europe is critical to empower our customers and to ensure AI innovation and autonomy remain at the heart of Europe." Mistral is also looking to recruit, with plans to expand the robotics team. Currently it is aiming to hire additional research scientists and engineers. Don't miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic's digest of need-to-know sci-tech news.

Mistral launches first robotics model in physical AI push

Paris-based Mistral AI has launched its first robotics model for industrial automation. This new system enables robot navigation using only a single camera. The Robostral Navigate model does not require advanced sensors or multiple cameras. It is designed to work with robots from various suppliers. This launch follows Mistral's recent acquisition of Emmi AI. Paris-based Mistral AI on Wednesday unveiled its first robotics model as Europe's leading AI company pushes into ⁠factories, warehouses ⁠and industrial automation. The launch follows Mistral's acquisition of Austria's Emmi AI in May and comes months ⁠after Paris-based startup Genesis AI unveiled a broader ⁠robotics model with navigation and manipulation capabilities.

Mistral Introduces Robotics AI That Requires Only One Camera | PYMNTS.com

The new Robostral Navigate is an 8 billion-parameter (8B) model and uses a single standard color camera (RGB camera), the release said. It can follow plain-language instructions such as "Leave the lobby, walk through the corridor, enter the supply room, and stop to face the second shelf." "To perform such tasks, other models often employ depth sensors, LiDAR, or several cameras working together," the release said. "Robostral Navigate uses only one ordinary RGB camera and no depth sensors, yet still achieves 76.6% on R2R-CE (room-to-room in continuous environments) validation unseen, the benchmark for following instructions in environments held out of training." The model enables robots to autonomously navigate offices, residential and commercial buildings, outdoor settings and other complex environments. It is likely to be used in manufacturing, delivery, logistics and hospitality, according to the release. "Robostral Navigate is only the first step toward a unified embodied agent," the release said. "We believe navigation is a foundational capability for general-purpose robotics. By combining large-scale simulation, efficient training and strong grounding priors, Robostral Navigate demonstrates that state-of-the-art embodied navigation can be achieved with a compact model and a single RGB camera." PYMNTS reported in March that physical AI and vertical AI startups are drawing investor attention as venture funding flows toward companies building systems designed to either operate in the physical world or automate specialized industry workflows. The surge in investment is visible across startups building machines designed to perform complex physical tasks, and these capabilities are becoming more commercially viable due to advances in AI models that combine computer vision, reinforcement learning and real-time planning, according to the report. It was reported in June that Mistral is in talks to raise about 3 billion euros (about $3.42 billion) at a valuation of about 20 billion euros (about $23 billion). Mistral was valued at 11.7 billion euros (about $13.4 billion) in a September Series C funding round in which it raised 1.7 billion euros (about $1.9 billion) and said it would use the new funding to fuel its scientific research. For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.

Mistral AI unveils robotics model for industrial navigation By Investing.com

Investing.com - Mistral AI announced a robotics navigation model Wednesday as the French startup expands into physical artificial intelligence following deals with major European industrial customers. The model, called Robostral Navigate, enables robots to navigate complex environments using a single camera and basic language prompts. The 8B model achieves 76.6% success on unseen R2R-CE benchmarks, outperforming multi-sensor approaches, according to the company. The model is hardware agnostic and was trained entirely through simulation. Robostral Navigate operates from a single RGB camera without LiDAR or depth sensors, beating the best single-camera approach by 9.7 points and the best system using depth or multiple cameras by 4.5 points. The model runs on wheeled, legged and flying robots and generalizes across robot sizes. The technology addresses applications across manufacturing, delivery, logistics and hospitality. The model combines pointing-based navigation with reinforcement learning, achieving a 79.4% success rate on validation seen and 76.6% on validation unseen. In May, Mistral said it struck deals with Airbus SE and BMW AG as it expands into advanced manufacturing. The Paris-based startup, founded in 2023 by researchers from Google DeepMind and Meta Platforms Inc. (NASDAQ:META), is in talks to raise around €3 billion at a €20 billion valuation, Bloomberg News reported last month. This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.

Mistral launches first robotics model in physical AI push

July 8 (Reuters) - Paris-based Mistral AI on Wednesday unveiled its first robotics model as Europe's leading AI company pushes into factories, warehouses and industrial automation. The launch follows Mistral's acquisition of Austria's Emmi AI in May and comes months after Paris-based startup Genesis AI unveiled a broader robotics model with navigation and manipulation capabilities. o Robostral Navigate enables robot navigation using a single camera o System does not require lidar, advanced sensors or multiple-camera setups o Designed to work with robots from different suppliers o Focused on navigation rather than object handling or manipulation (Reporting by Leo Marchandon in Gdansk; Editing by Matt Scuffham)