TLDR: VERSES AI Inc. has announced a significant advancement in robotics with its new multi-agent model, which reportedly outperforms existing methods on Meta’s Habitat Benchmark without requiring extensive pre-training. This innovation promises to enable robots to adapt and perform complex tasks in dynamic, unfamiliar environments more efficiently, potentially transforming automation across various industries.
VERSES AI Inc. has unveiled a revolutionary multi-agent robotics model that operates effectively without the need for pre-training, a common and often costly requirement in traditional robotics and AI systems. This breakthrough marks a potential paradigm shift in how robots learn and adapt, promising more flexible and reliable automation across diverse sectors.
The Vancouver, British Columbia-based company, founded in 2020, describes itself as a ‘cognitive computing company building next-generation agentic software systems’ inspired by natural principles. Its flagship product, Genius, is designed around first principles from science, physics, and biology, enabling it to generate reliable predictions and decisions even under uncertain conditions.
Hari Thiruvengada, Chief Technology Officer of VERSES, emphasized the significance of this development, stating, ‘I believe that by combining our world modeling and our active inference capabilities, we’ve shown robots can think on their ‘feet’ — navigating and completing complex tasks without months of costly training.’ He added that this innovation ‘has the potential to transform how robots operate across industries, from factories and warehouses to homes and public spaces, potentially unlocking a new era of truly adaptive, reliable automation.’
Traditional robotics often falls into two categories: ‘drive-by-wire’ systems, which are entirely pre-programmed and rigid, and ‘deep learning’ approaches, which, while more flexible, demand vast amounts of training data. VERSES highlights that both methods struggle with adaptability; drive-by-wire robots can halt at minor unexpected obstacles, while deep learning models, despite extensive training, may still falter with subtle environmental changes, such as a misplaced chair or a fallen bottle.
VERSES’ new model addresses these limitations by allowing robots to adapt through environmental exploration rather than relying on pre-trained data. The system integrates vision, planning, and control modules, enabling robots to autonomously handle unforeseen obstacles or retrieve dropped items. The company draws an analogy to human learning: ‘When a human needs to get a drink in a new apartment, they don’t execute by having practiced this task in hundreds of different apartments; they are able to adapt because they have a model of how the world works.’
Research conducted by VERSES’ lab, detailed in their paper ‘Mobile Manipulation with Active Inference for Long-Horizon Rearrangement Tasks,’ compared their model against a deep learning alternative across three household tasks: tidying a room, preparing groceries, and setting a table. The VERSES robotics model achieved a success rate of 66.5% across these tasks, outperforming the previous best alternative’s 54.7% success rate. Crucially, the VERSES model required no training, whereas the ‘multi-skill mobile manipulation’ baseline model necessitated 1.3 billion steps of pre-training for several skills across the tasks.
Sean Wallingford, former president and CEO of Swisslog, a prominent logistics automation company, commented on the implications: ‘Currently, robotics systems are often brittle and need huge amounts of training data, which makes them expensive and prone to going wrong.’ He praised VERSES’ approach, noting, ‘If we can deploy robots without training, they will be viable in a wide range of activities, from factories and warehouses to domestic and commercial applications.’
Also Read:
- Agentic AI Startups Secure $2.8 Billion in Early 2025 Venture Capital, Signaling Transformative Workplace Shift
- Microsoft Revolutionizes Platform Engineering with Advanced AI Agents
This development from VERSES AI signals a significant step towards more robust, adaptable, and cost-effective robotic deployments, potentially accelerating the integration of intelligent automation into everyday environments.
