Vsim, a Cambridge-based robotics startup founded by Michelle Lu and Kier Storey, has created a training system that rapidly prepares robots for real-world tasks. The company’s approach uses virtual simulation environments where tasks can be performed millions of times in computer simulations to identify optimal solutions before uploading them to physical robots. This process takes only minutes, substantially faster than rival systems.
The startup optimized its software specifically for graphics processing units (GPUs) used in artificial intelligence, moving beyond algorithms that date back several decades. The resulting system is efficient enough to run on the hardware carried by robots themselves, enabling machines to simulate tens of thousands of scenarios while moving. This capability allows robots to anticipate potential outcomes and adapt to unexpected events in unstructured environments such as homes and offices.
Vsim operates as a 10-person engineering team competing in a market dominated by much larger players like Nvidia, which employs hundreds of engineers in its robotics division. Nvidia offers a suite of software tools including simulation training systems and a world model called Cosmos that helps robots understand physics and environmental changes. However, even well-resourced companies face challenges in creating accurate virtual representations of the real world, particularly for complex manipulation tasks.
Researchers outside the industry acknowledge both the promise and limitations of simulation-based training. Rika Antonova, an associate professor at Cambridge’s Department of Computer Science and Technology, notes that while rapid simulation is promising, virtual environments remain imperfect approximations of reality. Certain physical phenomena, including deformable objects and cutting, remain difficult to model accurately in simulation. Both Nvidia and Vsim are actively working to reduce these gaps and improve the fidelity between simulated and real-world performance.
Article Attribution | Read More at Article Source
Article summary produced by Claude AI