AMR Challenges in Safety

AMR safety is often one of the first questions asked by potential customers — and rightfully so. Vehicles in industry settings have historically had a spotty safety record. For an uninformed or under-informed customer, when the word “autonomous” is attached to the word “vehicle,” instinct presumes a rise in safety threat.

In reality, AMRs should be relatively safe: ANSI (American National Standards Institute) and ISO have both published safety standards and requirements. AMR manufacturers are themselves extra diligent, applying safety precautions that go beyond adherence to safety standards.

AMRs, however, still suffer from safety issues, which will keep growing as manufacturing and logistics facilities change.

The Drawbacks Of Current AMR Safety Standards

AMR safety is commonly divided into two safety zones: the first is within the immediate perimeter of the AMR, where bumper sensors or pushing an E-stop switch stop the vehicle.

The second safety zone is the three-to-five feet perimeter: sensors such as RGBs and LIDARs, identify a moving object that blocks the AMR’s path and bring the AMR to a stop. Passive safety systems are also included, such as warning lights or Audible Warning/Alarm Signals.

The two-zone safety method, however, has its limitations. AMRs have tunnel vision -their eyes and ears are embedded in the robot — so their ability to map and understand their environment is fairly limited. Even if AMRs did have multi vision, just like humans, in an industrial environment it is simply not enough.

An Interesting paper published by an Engineering Laboratory at NIST (National Institute of Standards and Technology) states that vehicle-braking systems “cannot be expected to function as designed should an object suddenly appear in the path of the vehicle…” The paper goes on to say that current safety methods, namely 2D LIDAR sensors, are insufficient for what it calls “non-line-of-sight conditions” and concludes that “next generation agile manufacturing facilities, with their changing vehicle paths and human/AGV collaboration, may not allow dedicated vehicle paths and therefore increase the danger to humans from vehicles.”

The Need to Recalibrate the Mobile Robot Evolution

So how do we increase AMRs safety? The NIST paper proposes adding more safety measures such as 3D sensors, and requiring workers to wear tracking devices such as RFID tags (so AMRs can spot them even if they’re obscured of sight).

Yet those added precautions, as the NIST paper itself points out, are far from adequate. They also turn a pricey machine (already out of the reach of most warehouses and factories) into an even pricier one.

Here is where we need to stop and ask: is this the evolution path we envisioned? AMRs were supposed to replace the older AGV (Automated Guided Vehicle) model with its expensive and complex infrastructure. So what did we achieve by replacing them with other expensive machines that, yes, propose greater autonomy to some extent, but are complex to set up and deploy, require more engineers to babysit them and are unequipped to deal safely with the next generation of agile manufacturing?

If full autonomy comes at a price of compromised safety and/or higher costs, perhaps we are not on the right path?

The Hazards Of Deploying Multiple Types of AMRs

To This complex picture we add another complexity: deploying multiple types of robots on the same floor. For example, autonomous cleaning and for material handling. How do you make sure all robots work in unison?

This may bring to mind two robots colliding (like Star Wars’ R2D2 and his epic robot fight with R3F6). The reality today is not as dramatic. It’s usually about the two robot fleets not completing their work as planned.

However, in highly populated areas and agile manufacturing floors, this brings up problems, not least of which is safety: how do you make sure that multiple AMRs operating on a floor populated with employees do not bump into each other or harm the employees?

So why are Current MRs relatively safe?

Before We dive into the solution part, we need to answer this question, and it’s simple: AMRs are relatively safe because they are really slow. Remember that spunky Seat video we talked about in our last blog post that showcases AMRs in a car factory? Well, It’s shot in speed motion — as are most AMR marketing videos out there. In reality, almost all AMRS are inching across at sloth speed of about 1–3 km/hour.

This safety comes with a price tag — literally. Slow AMRs means lower throughput and lower productivity, more AMRs for the same job, which costs more money. In our previous post, we addressed the price challenge of deploying AMRs. As we just saw, safety is to a large extent an issue of price with just about anything — including AMRs.

The Solution

MusashiAI (MAI) introduces a new autonomous driving technology concept for industrial mobile robots, adding a new layer of safety to the existing two layers, by moving the “eyes” and “ears” of the AMR to the facility ceiling, and it’s navigation and management “brain” to an AI-based central ‘control tower’ system.

Unlike the existing AMR tunnel vision, MAI’s patented control tower architecture enables to see everything on the floor beyond the prism of the AMR (the so-called, non-line-of-sight conditions).

The cameras continuously map and monitor the floor, and the central system “brain” analyzes the location, trajectory, and speed of every object; classifying obstacles as either static or dynamic, and making on-the-fly decisions to either continue on the same path, make a detour and redirect, or in a case of a real emergency — stop.

The MAI control center changes AMRs paths or alerts people/objects long before they come within five feet of the AMR. Problems can be anticipated before they take place, and using machine learning the control tower can optimize routes according to shifting work patterns on the floor. This way the AMR is truly in synch with the dynamic work floor.

Orchestrating the Work of Multiple AMR Types and People

The MAI system can navigate and optimize the work of multiple types of AMRs operating simultaneously on the same floor — all through one central system software.

Our technology is platform-agnostic so it can control any type of an electric platform and drive it autonomously. It can also turn a an electrical pallet jack into a fully autonomous mobile robot, bringing autonomous vehicles within reach of many warehouses, logistical centers and manufacturing facilities that until now could not afford AMRs.

A New Autonomous Robot Paradigm

Our solution is more than an added layer of safety. What we are offering is a new paradigm of industrial robotics.

We are substituting the AMR (in effect an insular robot), which is prohibitively costly and lacks the aptitude and safety level required by the new dynamic factory.

We are offering a model of a vehicle that uses infrastructure, but one that is negligible in cost, quick and easy to install, and enables the vehicle to roam freely on the floor, to work in perfect tandem with other automated vehicles and people to achieve optimal productivity and safety…

Contact us to learn more before planning your next AMR fleet.

Maestro@634.AI