At a Stellantis assembly plant in Detroit, a long-standing challenge with industrial robots is being addressed through new vision technology combined with artificial intelligence. Traditionally, many factory robots operate "blind," programmed to perform repetitive tasks without the ability to perceive their surroundings. This limitation often leads to errors and costly downtime when parts are slightly out of position or racks age and deteriorate.

One particular robot arm, responsible for picking vehicle panels for welding onto Dodge Durangos and Jeep Cherokees, frequently froze or crashed when panels were misaligned. These interruptions could last up to an hour, causing significant bottlenecks on the production floor. To overcome this, Stellantis partnered with Inbolt, a French start-up specializing in adding vision capabilities to existing industrial robots.

Inbolt’s system features a compact camera and specialized software mounted on the robot’s wrist, enabling it to "see" and identify parts on the rack. The system uses 3-D mapping and computer-aided design (CAD) models to guide the robot’s movements, allowing it to adjust in real time to variations in part placement. Managers at Stellantis reported that within weeks of installation, the robot began operating smoothly, reducing downtime and enabling workers who had previously monitored the machine to be reassigned.

The upgrade was particularly notable because it was implemented on aging Fanuc robot arms, some of which had been in use since 2010. Typically, integrating modern vision technology requires costly new machinery, but Inbolt’s solution works with older equipment, providing a more affordable and scalable upgrade path. Since the initial deployment in Detroit, the technology has been installed on other robots in the plant, including one that applies sealant to car bodies and previously suffered frequent nozzle breakages.

Industry experts highlight that most of the world’s approximately 4.7 million industrial robots remain "blind," with only a small percentage equipped with vision systems. However, advancements in affordable 3-D cameras and AI are changing that landscape, allowing robots to handle more complex and variable tasks autonomously. Aaron Prather, director of market intelligence at a robotics industry association, said this shift is among the fastest-growing trends in automation.

Inbolt was founded in 2019 by graduate students from French institutions who initially designed a system to help human workers operate machines more efficiently. After exposure to global robotics developments, particularly in China, they pivoted to focus on enabling robots with vision. Their software, paired with relatively inexpensive off-the-shelf cameras from RealSense, distinguishes their offering by enabling precise 3-D environment mapping and part recognition.

Since its founding, Inbolt has raised roughly $30 million and deployed its vision system on over 200 robots globally, including a recent installation at a Stellantis plant in Hungary. According to Stellantis managers, the addition of vision technology has contributed to making the Detroit plant a top performer within the company. Anecdotal evidence suggests that improved reliability and reduced downtime could translate into significant cost savings, given that factory stoppages can cost thousands of dollars per minute.

Despite the benefits, challenges remain. For example, last summer a cable connecting a robot arm’s camera broke, halting production until the issue was quickly resolved. Nevertheless, Stellantis officials continue to expand the use of visual AI systems, underscoring the growing role of advanced perception technologies in modern manufacturing.