Eka Robotics, a startup founded by MIT professor Pulkit Agrawal and former Google DeepMind researcher Tuomas Haarnoja, has reached a valuation of $1.45 billion following a $150 million Series A funding round led by a16z and F-Prime. The company is generating excitement in the robotics industry with its innovative robotic arm, which combines a force-sensitive gripper with a unique deep-learning model focused on physical interaction. This model allows the robot to pick up a diverse range of objects with remarkable speed and precision, outperforming humans by handling roughly 1,500 objects per hour with a success rate exceeding 95%.

What sets Eka apart is its emphasis on force feedback rather than relying solely on vision-based AI models. Unlike traditional approaches that focus primarily on visual cues, Eka integrates a vision-force-action strategy, using physical force sensing to guide manipulation tasks. This method is key to its ability to handle delicate items, like raspberries, without crushing them, mimicking human-like dexterity. The company uses physics simulations to train its reinforcement learning system at scale, producing thousands of hours of virtual robotic experience rapidly and allowing the robot to exceed human demonstrations in performance.

Industry experts have expressed strong admiration for Eka’s technology, describing its movements as smooth, adaptive, and surprisingly human-like, with a novel ability to retry tasks multiple times if initially unsuccessful. This level of dexterity and reliability in manipulation tasks is a significant advance, as most robots struggle with such precision and adaptability. Despite the impressive demonstrations, Eka is still conducting pilot projects in industrial environments and faces challenges transitioning from lab conditions to real-world manufacturing lines, where durability and long-term reliability are critical.

Eka’s founders believe that their bottom-up approach—focusing on low-level physical intelligence first—enables robots to be more generally capable and ready to integrate with higher-level planning AI systems. While the robotic gripper hardware is cost-effective and designed for millions of cycles, Eka’s immediate focus is on excelling in particular verticals like pick-and-place and wire handling. This pragmatic path aims to prove the technology’s value in industry applications before expanding its capabilities, potentially transforming how robots handle physical tasks in the future.