Inbolt, a pioneer in real-time vision-guidance systems for industrial robots, has launched a new bin picking solution that brings human-like adaptability to manufacturing environments. The advanced system, powered by on-arm AI vision, enables robots to identify, grasp, and place parts in unstructured settings with 95% success rates and sub-1-second pick cycles—demonstrated in live automotive production. The solution marks a major shift from traditional bin picking methods, which depend on fixed, high-end overhead 3D cameras, complex calibration, and pre-programmed grasp points. These systems are costly, inflexible, and often fail when parts are misaligned, partially obscured, or bins shift during operation. Inbolt’s approach redefines the process by mounting a 3D camera directly on the robot arm, allowing the system to continuously perceive, understand, and adjust in real time. Leveraging Inbolt’s proprietary AI, the robot can generate an infinite number of grasp strategies on the fly, eliminating the need for a perfect or pre-determined grip. This closed-loop system enables the robot to see, attempt, and adjust its motion instantly—mimicking the way a human worker would adapt to variable conditions. The on-arm camera architecture significantly reduces hardware costs and setup complexity. Unlike traditional systems that require multiple fixed cameras and high-resolution sensors, Inbolt’s design allows a single robot to handle different bins, part types, and configurations with minimal reconfiguration. This flexibility makes it ideal for dynamic production lines and rapid changeovers. Albane Dersy, COO of Inbolt, emphasized the system’s real-world impact: “Traditional bin picking systems are too rigid for real factory conditions. We designed our solution to adapt in real time, able to see, grasp, and adjust the way a human would. That level of flexibility is what manufacturers need to reach truly autonomous production.” The system is already in use across more than five manufacturing plants, delivering consistent uptime and high throughput. It runs on NVIDIA’s hardware platform and uses Inbolt’s AI models for real-time pose estimation and continuous trajectory correction, ensuring reliable performance across diverse part geometries and use cases. Key advantages include faster deployment, lower total cost of ownership, and greater resilience in unstructured environments. The technology is compatible with major robot brands including FANUC, ABB, KUKA, Yaskawa, and Universal Robots, and can be trained on CAD models in minutes. Trusted by global manufacturers such as Stellantis, Toyota, Beko, and Volkswagen, Inbolt’s solution powers over 70 factories worldwide. The company’s generalist AI model allows any robot to master any task, any part, and any station—enabling faster automation rollouts, reduced downtime, and increased production efficiency. Manufacturers interested in modernizing their bin picking operations can request a live demo or pilot deployment at www.inbolt.com/contact.