The world of delivery robots is an exciting yet challenging frontier, and recent incidents in Chicago have brought this developing technology into the spotlight. Personally, I find it fascinating how these robots, despite their advanced capabilities, still face significant hurdles when it comes to navigating our complex and ever-changing urban environments.
The Growing Pains of Delivery Robots
Delivery robots, much like self-driving cars, encounter "edge cases" - situations that software developers and simulators haven't fully anticipated. As Vijay Kumar, Dean of Engineering at the University of Pennsylvania, puts it, "There are millions of things that can go wrong and will go wrong." This unpredictability is a major challenge for developers, especially in an environment as dynamic and unstructured as our cities.
The March Collisions
In March 2026, two delivery robots, one from Serve Robotics and another from Coco Robotics, collided with CTA bus shelters in Chicago. These incidents went viral, sparking concern and debate about the safety and efficiency of these robots. What makes this particularly fascinating is the variety of factors that contributed to these collisions.
In the first incident, three internal sensors on the Serve robot failed simultaneously, making it difficult for the robot to detect the bus shelter's glass. This highlights the complexity of the robot's internal systems and the need for robust redundancy measures.
The second collision involved a combination of environmental conditions and human error. Muted lighting due to weather conditions made the bus shelter glass difficult to detect, and a remote operator's decision to navigate through the shelter led to the crash. This incident raises questions about the role of human operators and their judgment in these situations.
Visual Challenges and Seasonal Variations
One detail that I find especially interesting is the impact of seasonal variations on the robots' ability to detect glass. Bus shelters that collect dirt and snow in winter become more detectable, while transparent glass after a rainstorm can be harder for the robots to see. This is a perfect example of how these robots must adapt to the changing conditions of our environment.
Malakhi Hopkins, a PhD student working on visual sensors, is developing solutions to help robots better judge the distance and angle of mirror-like surfaces. Additionally, an ultrasonic sensor that uses sound to detect clear glass is in the works. These innovations demonstrate the ongoing efforts to improve the robots' perception and navigation capabilities.
Learning from Mistakes
After the collisions, both companies took steps to improve their robots' performance. Serve Robotics implemented software updates to slow the robots down around bus shelters, and Coco Robotics reinforced protocols with additional operator training. These responses show the industry's commitment to learning from mistakes and continuously improving safety.
The Future of Delivery Robots
The delivery robot program in Chicago is set to continue until May 2027, and the city has established a 311 category for residents to submit their concerns and feedback. This engagement with the public is crucial for the successful integration of these robots into our communities.
In my opinion, the challenges faced by delivery robots are a natural part of technological development. As these robots continue to learn and adapt, they will become more efficient and safer. The incidents in Chicago serve as valuable lessons, pushing the industry forward and highlighting the importance of ongoing research and development.
