By "Autonomous Mobility by Shared Autonomy," I meant, is a technology stack that based on driving contexts, automatically and seamlessly switches between different modes of driving: autonomous and remote driving.
For the past two years, I ran my own startup, Atlas Robotics, Inc., and worked on developing a technology stack for autonomous mobility by shared autonomy. The idea of ``Autonomous Mobility by Shared Autonomy'' is driven by a technology stack that based on the complexity of driving contexts, automatically and seamlessly switch its autonomy back and forth between two driving modes: autonomous and remote. "AtLast," the product created from this concept will enable a self-driving capable vehicle to autonomously navigate less complex navigation scenarios like driving on a predetermined route while occasionally interacting with road occupants (e.g., vehicles, pedestrians, etc.), and allows a human operator at a remote site to fully or partially take control of the vehicle once its navigation limit is surpassed by the complexity of the environment. We were building such a hybrid driving system because we strongly believe that the benefits of autonomous driving are too substantial and urgent to be ignored or delayed, and inherent challenges to complete fully autonomous driving are too complex to overcome in a couple of decades. This is not just our own assessment of the development of autonomous driving technology but also the consensus of the community leaders. ("...The technology won't be ubiquitous for decades and that driverless vehicles will always have constraints...", John Krafcik, CEO, Waymo, Bloomberg, Nov 13, 2018; "...This is a change that's going to scale over decades, not over weeks...", Chris Urmson, CEO, Aurora, Mar 25, 2019; "Nobody is even close" to making a truly self-driving car, James Kuffner, CEO, Advanced Development Division of TRI, Oct 30, 2019.)
Our hybrid driving system aided by human-in-the-loop will be a stepping stone before the community appreciates the benefits of fully autonomous driving everywhere. The utmost utility of our hybrid navigation system becomes apparent when a remote operator can drive more than one robotic vehicle at a time. This desideratum requires our system, one that enables entities (e.g., autonomous driving SW and human remote operator) to seamlessly switch the autonomy of the vehicle when necessary. In turn, to support seamless transition, the shared autonomy module should reliably execute the following tasks:
The target customers are those businesses who need workers with specific skillsets. Examples include those providing shuttle services, large-scale farming agriculture, premise patrol, closed-space transport, and construction services. These customers are all suffering a lack of manpower and the highest liability insurance premiums. To remedy these pains, they all need something that can do what a skillful worker can do, without exposing workers to any hazardous situations. AtLast can relieve these pains by filling the gap of manpower shortages and/or eliminating worker exposure to any hazardous work situations. For example, the developing system will enable law- and security-enforcing of officers to more safely and effectively patrol regions without exposing themselves to dangerous situations; a shuttle driver, an operator of construction or agricultural equipment does not have to be behind the wheel but at a remote station, where they could potentially drive more than one vehicle at the same time. When something requiring human attention happens, the developing algorithm notifies the operators to take over and lets what a human does better. For instance, the developing algorithm will switch an excavator driven autonomously to a construction site to a human operator at the remote station when the operator needs to carry out precision earth digging. This idea can be also easily projected onto closed-space transport domains such as shuttle services for silver communities and freight delivery for airports where an operator can supervise autonomous operations of multiple vehicles.
Atlas Robotics, Inc. was established in May 2018 envisioning the development of robots to empower humans. We were a team of five full-time engineers working at two offices, Livermore, CA, and Seoul, Korea. With a seed fund of $1 Millions, we developed two navigation systems for autonomous driving and for remote driving. See the following as a list of the developed technologies. For customer traction, we were in conversations with a couple of potential customers -- unfortunately none of these led to anything tangible by the time we decided to pull the plug. The followings are the IPs we developed and obtained:
IP | Description and Demo Videos |
---|---|
Autonomous navigation stack | Our vehicle is autonomously driving on a predetermined route while localizing itself with respect to the map we built. |
Lidar SLAM | Building a map of the route about my commute from the office to home. |
Building a map of the Bunker City at GoMentum station. | |
Building a map around the office. | |
Visual SLAM | Building a map of the open space at the front of SNU BI. |
Validating our visual mapping stack against the KITTI sequence 08. | |
Remote driving | Demonstrating a SW stack that enables an operator to remotely drive a vehicle over LTE network, thru a construction zone and a corridor between buildings. |
Another demo of remote driving. The packets of twist commands are encrypted by AES 128 before being transmitted to the vehicle. | |
US patent | Automated Transfer of Vehicle Control for Autonomous Driving (US 10,717,448 B1) |
I would like to take this opportunity to sincerely thank those who contributed to what Atlas Robotics Inc. was by the time we decided to shut it down -- we failed to raise funding to continue operating, partially due to COVID-19. As we couldn't pull it off (Yes, yes, Mea Culpa, but was it a flop? Well, it remains to be seen.), I won't name names here. But I want to say it clear and loud: None of the above mentioned would exist without your dedication and hard work.
Each and every one of you, Thank You! You did a great job!
p.s.: I'm glad to know if you find this intriguing. But remember what I talked about here is the results of two-years' blood, sweat and tears (you know, the part about "tears" is a figure of speech). So please don't just copy it and proclaim this is your idea. Instead let us have a chat, so send me an email. Another thing you'll have to remember is that everything here is protected by the U.S. and International Copyright Laws and some of it are also protected by a U.S. Patent. ©2021 Youngwoo Seo, All Rights Reserved.