In 1986, the first self-checkout system was introduced at a Kroger grocery store just outside Atlanta. After decades, the expertise has finally spread nationwide across the United States. As the retail landscape increasingly relies on automation, the inevitability of robotic bagging in grocery stores becomes more apparent.
MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) this week is showcasing innovative technologies. The system seamlessly integrates computer vision with a sleek robotic gripper to efficiently package a wide range of diverse items. Researchers positioned ten unfamiliar objects on a grocery conveyor belt to test the system’s capabilities.
The assortment of goods spanned a wide spectrum, featuring fragile items such as fresh produce – including grapes, bread, kale, and snacks like muffins and crackers – alongside more substantial commodities like canned soups, meal containers, and ice cream tubs. The imaginative and prescient system’s initial activation involves object detection, with subsequent processing focused on measuring and orienting these objects along the belt.
As the fingers brush against the grapes, sensitive sensors within detect their fragility, prompting a conclusion that they shouldn’t be placed at risk by being pushed to the back of the bag – a lesson many have learned the hard way? The subsequent observation highlights that the soup can’s rigid design causes it to become lodged at the bottom of the bag.
According to Annan Zhang, a lead author on the study, this breakthrough represents an essential initial step towards realizing robots’ potential to pack groceries and other items in practical settings. “While we may still need to refine our readiness for commercial implementation, our findings unequivocally showcase the potential benefits of seamlessly integrating multiple sensing modalities within advanced robotics systems.”
Despite notable progress, there remains ample opportunity for further enhancement, particularly with regards to upgrading the gripper and imaging system to better determine the optimal packing sequence. Because the system has been upgraded to extraordinary strength, it could potentially extend its operations beyond the confines of a typical grocery store into more industrial settings such as recycling facilities.
