Ottonomy offers Contextual AI 2.0, putting VLMs on the edge for robots
Ottobots use Contextual AI 2.0 with embodied VLMs in edge robotics. Source: Ottonomy
Ottonomy Inc., a provider of autonomous delivery robots, today announced its Contextual AI 2.0, which uses vision language models, or VLMs, on Ambarella Inc.’s N1 edge computing hardware. The company said at CES that its Ottobots can now make more contextually aware decisions and exhibit intelligent behaviors, marking a significant step towards generalized robot intelligence.
“The integration of Ottonomy’s Contextual AI 2.0 with Ambarella’s advanced N1 Family of SoCs [systems on chips] marks a pivotal moment in the evolution of autonomous robotics,” stated Amit Badlani, director of generative AI and robotics at Ambarella. “By combining edge AI performance with the transformative potential of VLMs, we’re enabling robots to process and act on complex real-world data in real time.”
Ambarella’s single SoC supports up to 34 B-Parameters multi-modal large language models (LLMs) with low power consumption. Its new N1-655 edge GenAI SoC provides on-chip decode of 12x simultaneous 1080p30 video streams, while concurrently processing that video and running multiple, multimodal VLMs and traditional convolutional neural networks (CNNs).
Stanford University students used Solo Server to deliver fast, reliable, and fine-tuned artificial intelligence directly on the edge. This helped to deploy VLMs and depth models for environment processing, explained Ottonomy.
Contextual AI 2.0 helps robots comprehend environments
Contextual AI 2.0 promises to revolutionize robot perception, decision making, and behavior, claimed Ottonomy. The company said the technology enables its delivery robots to not only detect objects, but also understand real-world complexities for additional context.
With situational awareness, Ottobots can better adapt to environments, operational domains, or even weather and lighting conditions, explained Ottonomy.
It added that the ability of robots to be contextually aware rather than rely on predesignated behaviors “is a big leap towards general intelligence for robotics.”
“LLMs on edge hardware is a game-changer for moving closer to general intelligence, and that’s where we plug in our behavior modules to use the deep context and adds to our Contextual AI engine,” said Ritukar Vijay, CEO of Ottonomy. He is speaking at 2:00 p.m. PT today at Mandalay Bay in Las Vegas.
Ottonomy sees numerous applications for VLMs
Ottonomy asserted that Contextual AI and modularity has been its “core fabric” as its SAE Level 4 autonomous ground robots deliver vaccines, test kits, e-commerce packages, and even spare parts in both indoor and outdoor environments to large manufacturing campuses.
The company noted that it has customers in healthcare, intralogistics, and last-mile delivery.
Santa Monica, Calif.-based Ottonomy said it is committed to developing innovative and sustainable technologies for delivering goods. The company said it it is scaling globally.
Register today to save 40% on conference passes!
The post Ottonomy offers Contextual AI 2.0, putting VLMs on the edge for robots appeared first on The Robot Report.