Industrial Perception, Inc. (IPI) is one Silicon Valley startup with its eyes on the demands of tomorrow’s commerce. The company is set to seriously alter robotics across the supply chain by pushing the limits of today’s vision-guided machinery. Their more human-like perception technology is ready for use in such industrial tasks as automated truck unloading, bin picking and parts feeding. IPI spun out of the famed Willow Garage institute last spring and houses some of the keenest brains in the business. Two of the company’s four founders, Dr. Kurt Konolige and Dr. Gary Bradski (who founded Open CV) happen to be two of the foremost authorities on vision technology in robotics.
The two met at Willow Garage and, over the past few years, have collaborated to construct software architecture capable of 3-D vision and (soon) navigation skills, that no one else in automation has been able to accomplish. Their technology equips existing industrial robots with the eyes and brains necessary to distinguish items from a pile of touching and overlapping objects (even if they are similar in color) and differentiate between different item surface depths with scale. In an industrial automation first, IPI robots use that information to make intelligent determinations about how to optimally grasp, reorient and guide those items through space, avoiding collisions in real-time.
A multi-dimensional future for robotics
3-D vision is what’s new and next for robot perception. The technology has been available for a while (at considerable expense), but without advanced programming for many commercial applications. When the Kinect came along, it made low-cost 3-D vision available to the robotics community, and became a new standard for research into robot vision. IPI similarly employs infrared (IR) dot technology to map the workspace, but IPI has optimized its software to operate with the robustness, safety and speed required for industrial environments. Current industrial robots, when they are equipped with vision-guided technology, labor under the limitations of two dimensional perception. When performing pick-and-place or palletizing operations, existing technology can recognize one object and pick it–assuming there is nothing else in the workspace–from a shallow bin of similar objects. The restrictions of the robot, therefore, prevent the introduction of broader bins or fuller trucks–the kinds of challenges which human workers can more easily address. Comparatively, using a combination of stereo cameras and software, an IPI-enabled robot becomes master of an entire sensing environment. The system will also interface with customer equipment to communicate state, failure modes and retry logic. IPI technology means robots can sort through deeper bins or unload trucks full of randomly-oriented objects, even scanning barcodes in the process; it means robots are much closer to human-level perception and information sharing at a rate beyond human capacity.
Starting with the basics
It’s easy to take for granted the simple logistical tasks that are vital to an efficient supply chain. Truck unloading is not the flashiest task for a robot to perform, but it is a serious problem area for warehouses and distribution centers where e-commerce demands faster turnaround and human workers risk suffering from serious stress injuries. The material handling space is where the robotics industry made its debut, and where the vast majority of profit within the now $25.5 billion dollar industrial robotics market still remains. Wynright Corporation, a material handling integrator, is one of IPI’s business partners and has identified the loading dock as a last frontier for automated logistics. That grueling workspace is where IPI aims to make its first splash, with demonstrations at Automate in robotic truck unloading, among other skills. For now, IPI is perfecting the kinds of simple industrial applications that can make a big difference.
What followers of this technology can expect to see going forward is the software behind today’s automated truck unloaders becoming the technology that guides tomorrow’s advanced assembly robots. As the company’s product manager, Erin Rapacki, notes, “Robotics won’t advance until companies are successful at doing these basic things first. This vision technology can do a lot of things, but it’s not going to get better until it’s a product somewhere. This is where robotics is right now.” In addition to Wynright, IPI has partnered with Yaskawa Motoman Robotics to refine vision guided tasks with their robotic arms, but is in open discussion with the industrial community and beta partners as the company progresses quickly from special projects towards a broad commercial product. Rapacki expects that transition to take place within a couple of years. Visit booth 365 at Automate in Chicago, Jan. 21-24 to see IPI’s robots at work.