What You Will Learn
Courier and delivery services are a $246B/yr industry with high personnel costs and pressure to perform on increasingly short timescales. Consumer interest in next- or same-day delivery has driven recent projects to dramatically reduce the delivery time for both local retailers and large national brands.
All of the currently available delivery services are based on some variant of human courier, whether these workers operate on a purely contract basis (as with Uber) or are employees of a traditional dedicated courier service (as with FedEx). Automation in the courier and delivery space has been applied in package sorting and regional distribution centers, as well as in route planning and schedule optimization, but has not made significant inroads into the actual handling and delivery of packages. The “last mile” has remained a task exclusive to human couriers.
The massive investment (~$150B to date) in autonomous vehicles will ultimately address part of the last mile problem. Most of the 46 corporations listed by CB Insights working on autonomous vehicles target package delivery as a key component of their overall value proposition. But autonomous vehicles can not solve the “curb to doorstep” problem. The package must either be stored in a locker or the customer has to come to the curb to receive the item.
We suggest that a fleet of autonomous vehicles working with curb-to-doorstep robot couriers will be the long-term solution for delivery. Aerial drones, being considered by some retailers for this purpose, have limitations in weight, safety, energy, and regulation. Wheeled robots, including those already implemented for autonomous delivery, have challenges navigating human environments such as stairs. While these solutions may solve a limited set of delivery environments and situations, they cannot be as general or capable as a human courier operating in human environments, which we have made conveniently navigable by legged locomotion.
Agility Robotics, building on a strong research background in physics of legged locomotion, designs and builds bipedal robots that can go where people go. We have demonstrated capable legged locomotion over most human-navigable terrain, including stairs. Our bipedal robot, Digit, incorporates arms that can pick up, carry, and set down packages; sensors that enable the robot to choose foot placement as it navigates rough terrain; and the toughness to handle accidental falls and other hard use, and keep going.
The flexibility of legged robots to move within human environments enables an almost unlimited number of markets to explore applications. We are focusing this capability on the final few meters between an autonomous vehicle and the customer’s delivery location, which offers the opportunity to address a vast market, while also giving the technology time to mature and grow to address other general tasks such as package handling and sorting. We are building the world’s most capable hardware for these tasks, and as the intelligence and perception and mapping capabilities of robotic systems continue to improve, will enable a broad set of use-cases for robots in human environments.
The science is ready; we know how legged locomotion works. The technology is ready; motors are powerful enough, computers fast enough, sensors that are sufficiently capable. It is time to aggressively begin use-case testing and evaluation, to identify the highest-utility applications, and work with partners who seek to gain the experience that enables good choices and effective solutions in implementing the final piece for autonomy in logistics, the last 50 feet of package delivery.