phonlamaiphoto - stock.adobe.com
Applications of autonomous robots lead in the enterprise
In enterprise AI, bot technology is leading the charge. Seamless integration and process streamlining are initial benefits, but the true profit lies in what comes next -- auto-AI.
Machines performing intelligent actions without human supervision is one of the long-time dreams of AI researchers, science fiction aficionados and futurists. A primary goal for artificial intelligence is to enable machines to do increasingly more complicated, human tasks. The primary value of autonomous systems is to minimize human labor, especially tasks that are dull, demeaning, dirty or expensive for humans to do.
While one might have robots or vehicles in mind when hearing about autonomous systems, there is a rise in self-operating software applications that are able to accomplish tasks, achieve goals, interact with their surroundings and perform their objectives with minimal human involvement.
Autonomous applications of AI have become so prominent across a wide range of application types and industries that it is one of the most adopted patterns of real-world AI usage. Enterprises have adopted the autonomous pattern for a variety of use cases, including automatic documentation and knowledge generation, autonomous business process and collaborative robots.
In a controlled environment, robots have been around for decades -- operating on factory floors, warehouses and in other locations to perform menial tasks. Robots are used increasingly to automate human labor; however, not all robots are intelligent, and many physical robots need to be caged off from human workers due to potential danger and risk.
In 1996, the collaborative robot (cobot) was born from university research projects and the General Motors Robotics center. Enterprise cobots are meant to operate in conjunction with and in close proximity to humans in order to perform their tasks. Humans provide the power while the cobots control and steer objects into place with precision.
Rather than being programmed on tasks to perform like traditional industrial robots, cobots are trained by demonstration. Humans control the bot by physically moving it around, and the cobot remembers the steps and perhaps even the end goal and then repeats those steps while optimizing them to achieve increasingly better outcomes. In this way, cobots are augmented intelligence bots that are meant to enhance but not replace humans and human tasks.
Since cobots can safely work next to humans, companies are finding many use cases for them. Cobots are now used for logistics and warehouse automation, delivery and mailroom activities, cleaning and customer interaction. Home improvement store Lowe's released their autonomous retail service cobot, LoweBot, in 2016 to roam store floors and help customers with simple questions, as well as assist with inventory monitoring on shelves.
In 2017, Walmart deployed shelf-scanning cobots in 50 locations around the U.S. These cobots autonomously roam store floors to check inventory, prices, spills or misplaced items.
Even museums are finding value from cobots. The Smithsonian Institution has started to use SoftBank Robotics' Pepper robot to add value to the visitor museum experience. Pepper is able to answer various questions relating to the museum or a specific exhibit, manage lines and help translate words and phrases into different languages.
Bots in the enterprise
Intelligent software bots are being used by many organizations to address back-office and enterprise systems-oriented tasks currently taking up much of the time of knowledge workers. Intelligent automation expands upon the ideas of process management, robotic process automation (RPA), content intelligence and enterprise integration to provide process awareness and improvement. In addition to simply automating tasks, intelligent software agents are autonomously discovering business processes and optimizing process activities.
RPA bots are useful, but not intelligent or fully autonomous. While RPA bots help with many of the repetitive tasks in the enterprise, without any real adaptive intelligence, these bots are merely automating mundane tasks. Many of these systems automate screen scraping, keyboard and mouse entry, using screen recorders, diagrammatic flow and process bots. A human still needs to be in the loop to handle process exceptions, changes in systems flows and incomplete or unstructured information.
By adding machine learning and intelligence, RPA systems gain the ability to autonomously discover and perform business processes in an agile manner. These intelligent autonomous software agents can then perform process discovery, analytics and optimization. The end result is systems that can operate without human supervision to run and optimize the businesses' back-end operations.
Autonomous software agents are able to extract information from various systems to generate documents and content. These systems are being used in heavily regulated industries to enhance compliance; increase operational efficiencies; or create customized repeatable documentation, such as legal papers, invoices or security and loan documents.
Enterprises use autonomous software agents to assist with internal support and service management. These bots can perform autonomous routing of internal tickets or documents to various departments and iteratively find better ways to handle process flow.
Forms of autonomous systems are finding use in supply chain and logistics applications, where they can plan and execute storage and transportation of goods and products from point of origin to the end point. Autonomous logistics improve shipping times and order fulfillment, provide accurate tracking of shipments and monitoring of events that could disrupt supply chain and increase accuracy of inventory forecasting to reduce under and overstocking.
AI is advancing the vision of cars driving themselves for everyday use closer to reality. Intelligent, self-driving systems have already been implemented in trains, boats and mining vehicles that operate in relatively controlled environments without a human driver. Train systems such as the Copenhagen Metro or the Barcelona Metro line 9 are able to operate on their own including moving from station to station, opening and closing doors, detecting obstacles and handling emergency situations.
While autonomous systems form a large part of the goal for AI implementations, focusing solely on autonomous vehicles as an ideal example of this pattern is misguided.
The Society of Automotive Engineers and the U.S. Department of Transportation's National Highway Traffic Safety Administration classify six levels of capability for autonomous cars, starting from completely human-operated Level 0 to fully autonomous Level 5.
Companies such as Tesla, Ford, Toyota, Uber, Volvo and others have been working very heavily toward Level 5 automation -- but after several years of activity -- there is still no commercially available vehicle operating even close to fully automated mode. While Tesla's Autopilot falls somewhere between Levels 2 and 3 and Cadillac Super Cruise operates at Level 2, well-publicized accidents and missteps show that even at this limited autonomy level, AI systems struggle in application.
Autonomous systems -- both physical and software forms -- are currently used by many organizations in varying complex capacities. As AI systems become even more advanced, companies will start to see fully autonomous systems moving physical goods as well as processing and moving information throughout their organization, finding bottlenecks and increasing efficiency and interacting with customers to improve the overall customer experience.