MOV.AI disrupts Autonomous Mobile Robot development with a Robotics Engine Platform that contains everything needed to quickly build, deploy and operate intelligent robots.
Have you ever dreamt about having a robot? I’m guessing most people involved in robotics have. For those of us who grew up in the ‘80s, the ideal robot was embodied by R2D2, the opinionated and loyal little droid that could do almost anything, including risking its very own existence. R2D2, and more recently, BB8 (ok, even 3CPO) embody the human desire to have a sidekick to help with the three Ds – Dull, Dirty, and Dangerous.
Fast forward to the modern day. Robotics has progressed by leaps and bounds. Putting aside sci-fi dreams, I doubt that 40 years ago there were many who could envision the impact that robots would have on manufacturing, health, logistics and so many other areas of our lives.
In reality, robots have parted from fantasy by moving away from personal assistance robots to commercial-use, task-oriented robots. This gap is naturally reflected in robot design that is optimized for the task the robots were designed to perform. But there’s an even more fundamental difference in the way real robots operate.
Robots today – even collaborative ones – are designed to perform a predefined task. They are mostly suitable for large-scale operations with well-defined processes and tasks. It’s no wonder that the vanguard of robot and AMR implementation is the automotive industry – the cradle of the production line, and uber operations such as Amazon.
It is in these settings that current cobots and AMRs thrive: Operations that use a large number of robots and require significant pre-planning and detailed definition routes and processes.
So what’s the problem? Humans simply don’t work that way. Most companies don’t work that way. The human brain does not operate in a linear fashion, meaning that automating work processes require plenty of effort and even adjustments. It’s no coincidence that less than 20% of warehouses today have some sort of automation scheme in place, and only 5% use sophisticated automation.
Add to that the fact that chaos and change are part of life. Malfunctions happen. Priorities change. Trends shift. Technology evolves. External factors affect internal processes and systems. While fully automated operations have many merits, there is one thing they are not, and that’s flexible.
The next stage in the evolution of robotics is allowing for chaos: enabling non-automated operation as part of automation, or hybrid automation. We all know that automation is a means to an end, not a goal in its own. When looking at the broader business goals and not just automation, it becomes clear that there’s more than one way to approach it, and that full automation is only a step in the evolution towards a more natural mode of operation.
Reality contains brownfield operations where full automation is either impossible or requires changing work processes and overhauling the facility. Reality contains mid-sized or small operations where a fleet of robots does not make economic sense. The real world is made up of young operations whose products, bill-of-materials and processes evolve on a daily basis. The real world contains people who have stopped to take a break, have forgotten a part or thought of a better way to perform a task.
The next generation of robots and automation systems needs to be able to not only perform predefined structured tasks, but to deal with unstructured tasks and ad hoc requests. They need to provide equal value in a highly organized logistics center and in a garage. To do that, they need to be able to perform ad hoc, changing, non-deterministic actions, based on human decisions. The goal is not automation, but to support humans in completing their tasks.
Let’s consider some potential hybrid automation use cases:
Cooperative robots in hybrid automation models require slightly different capabilities than those designed for fully automated environments. Such cooperation requires more advanced Artificial Intelligence (AI) and machine learning capabilities that allow robots to deal with non-repetitive tasks, ad-hoc instructions and an environment that is much more dynamic and not as well defined.
Let’s consider the following fully manual example:
A worker in a workstation requires a heavy part or component. Instead of wasting time by going to fetch it and potentially injuring themselves, they use the app on their workstation or type in a location or requested item into an interface on the robot itself. The robot goes to the designated location, where another employee loads the AMR and flags the item as loaded in the system. The AMR returns to the work station, where the issuer of the task takes the item and marks the task as complete.
The original request, the indication that the part has been placed, and the task completion all need to be entered and logged. This can be done on the robot itself, in workstations, or through a central management system.
Input type can range from pre-programmed tasks to coordinates or zoned areas and maps.
The interface itself can be touch screens, commands, or even voice commands.
Working in brownfield sites alongside humans requires state-of-the-art localization, navigation, and risk prevention. AMRs operating in such spaces require dynamic mapping and AI-based perception that will help them interact with people and other machines.
This could mean high accuracy SLAM localization, human intent analysis, and the ability of the fleet to work in a swarm model to address dynamic situations.
Hybrid operation unites two opposing approaches: automated fleet management which relies on detailed and structured tasks and processes, and ad-hoc requests that reside at the opposite end of the spectrum – single tasks defined on the spot.
Task prioritization is required both for fully manual operations for cases where several requests are made and in hybrid operations where ad hoc requests need to be reconciled and incorporated into the automatically generated tasks.
Not all activities in manufacturing or logistics can (or should) be automated. Whether it’s due to complexity, operation maturity, scale, or a desire to leave room for human innovation and creativity, there is a need for AMRs to perform non-automated tasks and cooperate with people in a hybrid model.
The next stage in the evolution of industrial and warehouse automation is allowing this more natural mode of operation. It may in fact even speed up their adoption.