Shared Mobility

Ford package delivery tests combine autonomous vehicles, bipedal robots

Ford package delivery tests combine autonomous vehicles, bipedal robots

By Steve Crowe
Ever since it introduced its first bipedal robot, Cassie, in early 2017, Agility Robotics has said package delivery is an ideal application. We now have our first look at how the Albany, Oregon-based company envisions its bipedal robots assisting last-mile delivery.
Agility and Ford Motor Company are developing a last-mile delivery solution that combines Ford’s autonomous vehicles and Agility’s Digit bipedal robot. This system is in the early stages of development, but an autonomous delivery test program is scheduled to begin in early 2020.
Agility CEO Damion Shelton told The Robot Report the first time the entire system was up and running was just two weeks ago. The companies started discussing a partnership in February 2018 and began work on this project about six months ago.
“This is a big deal for [Ford] to be making this type of move,” said Shelton. “And for us it’s a big deal to be making firm decisions about what we’re using the robots for.”
Here’s how it works. Digit, which can fold up tightly, rides in the back of Ford’s autonomous vehicles. Using a patent-pending lift designed by Agility, Digit can get itself in and out of the vehicle. The lift mechanism latches onto the same place you would have third row seats and goes under Digit’s “armpits” to lift the robot in and out of the vehicle.

Digit, which can carry boxes up to 40 lb. (18 kg), picks up the package, walks up the front steps, drops off the package and returns to the autonomous vehicle.
One of the first questions with these types of videos is the level of autonomy involved. For this demo, local tele-operation was used to trigger behaviors for Digit at appropriate times — such as when to pick up the box and when to climb the stairs. But the actual picking up of the box and climbing of the stairs were performed autonomously and run locally on Digit.
The only thing “fake” part of the video are the dotted lines that indicate an obstacle in Digit’s path. Shelton said Digit did not fall once during the 12 hours of shooting the video.
“I’m as surprised as anyone because there was a bug in the system leading up to the shoot that caused Digit to fall excitedly and repeatedly. We didn’t know it was fixed until the day of the shoot.”
“We will repeat the same demo under 100 percent full autonomy,” Shelton said. “We’ll press a button and the full demo executes with no human involvement whatsoever. We’ll be able to go to any neighborhood, acquire a base map of the neighborhood, and then deploy the car and robot without someone giving it hints. We anticipate doing that by the end of this year or early next year.”
Shelton said upgraded versions of Digit being developed will help enable full autonomy. The second-generation Digit is coming in mid-2019 and the third generation in the fall. “The sensors on the robot in the video – LiDAR and Intel RealSense – are functional but aren’t used for planning. We’ve been testing version two of Digit with this sensor stack and plan to use it for full autonomy.”
Shelton said it’s more important to start capturing real-world data rather than adhere to the notion that it’s full autonomy or bust.

Help from the cloud

One interesting tidbit is how Ford’s autonomous vehicles could help Digit navigate edge cases or more difficult environments.
“A self-driving vehicle is capable of creating a detailed map of the surrounding environment, so why not share that data with Digit instead of having it recreate the same type of information?” said Dr. Ken Washington, Vice President, Ford Research and Advanced Engineering, and Chief Technology Officer. “After all, both Digit and the self-driving car need to know where they are in the world, where they need to go and how to get there. When a self-driving vehicle brings Digit to its final destination, the vehicle can wirelessly deliver all the information it needs, including the best pathway to the front door.
“The car could even send that information into the cloud and request help from other systems to enable Digit to navigate, providing multiple levels of assistance that help keep the robot light and nimble. Digit’s light weight also helps ensure it has a long run time, which is essential for a self-driving delivery business that will be operating most of the day.”
There haven’t been many instances, but this isn’t the first time a self-driving vehicle and legged robot worked together for last-mile delivery. ANYbotics’ quadruped robot, for example, was used by Continental to demo last-mile delivery at CES 2019.

What’s next for Digit the delivery robot?

Shelton said once Digit is capable of full autonomy for this application, testing will be done in Albany, Oregon and Michigan, which is home to Ford and a challenging weather environment.
“We’re very careful to say we don’t need to solve or address the hardest 90 percent of delivery problems. The easiest deliveries can be done with wheeled vehicles. But others can’t be done with wheeled vehicles, like the delivery in the video,” Shelton said. “All of those are test environments we can test in. We’re challenging ourselves just to the edge of what we can do all the time.”
Ford believes driverless delivery has huge potential. It said the potential value of the market for robot ride-hailing and driverless delivery is $332 billion.
“Logistics is a large market. If we gain 1 percent of the market each year, that’s an unbelievable growth rate for a robotics company,” said Shelton.
Shelton was clear this is not a commercial service competing with the UPS’ and FedEx’s of the world. And that the Agility-Ford partnership will remain in medium to large-scale pilot mode for a while.
“This is not a one-and-done concept for us,” Shelton said. “This is very much what we thought we’d be using Digit for. It’s earlier than we thought, but the opportunity was there now to partner with Ford. So we’re excited.”

Exit mobile version