At first glance, high-capacity Autonomous Mobile Robots (AMRs) and Automated Guided Vehicles (AGVs) perform similar tasks. That is, they are capable of moving pallets and heavy payloads long distances. However, a robot is only as good as the efficiency gains it provides your organization. It turns out that it’s not what the robot does that matters, but how. Since navigation is the biggest factor affecting your solutions’ efficiency, we will take a closer look at what goes into efficient navigation.
First, let’s define our terms. Navigation is the ability to get to the right destination (or destinations) at the right time. Think about your last trip to your favorite home improvement store. You use navigation to get from the front door to the plumbing department and back to check out.
The three distinguishing navigation features that help achieve greater efficiencies are: obstacle avoidance, localization, and pallet handling. A combination of these features makes the AMR smarter, more flexible, and more efficient than traditional AGVs, and free workers to do their jobs without distraction.
Obstacle avoidance is term that refers to the ability for a robot to navigate freely in dynamic environments and tight spaces.
Path Following and Path Planning
To understand obstacle avoidance, we’ll first take a look at the differences between path following and path planning. Path Following is the way most traditional and advanced AGVs navigate. It works by setting a specific, predefined route throughout a facility. Then, the AGV follows that path as accurately as possible. There are several drawbacks to Path Following. For one, obstacles in the path of a stopped AGV must be manually cleared, causing frequent interruptions to workers. By comparison, an AMR will independently navigate around the obstacle and continue on its way.
Passing Moving Vehicles
Some advanced AGVs will navigate around an obstacle, which works well for an isolated obstacle in an aisle. However, when traveling in high traffic areas, AMRs travel smarter. Imagine you’re on a two-lane highway and there’s a truck moving slowly ahead of you. You can pull into the passing lane to pull ahead. It’s not until you get into the passing lane that you see there are 2 more vehicles in the travel lane up ahead. They key difference lies in how and when a self-driving forklift pulls back into the travel lane. An advanced AGV will get in front of the truck it just passed, just to realize that there is another vehicle in front of it, causing it to have to start the process all over again. An AMR, however, will intelligently decide whether to stay in the passing lane or to merge back into the travel lane based on what it can see rather than blindly returning to the path.
Another key difference in the way AMRs and AGVs handle obstacles is that obstacles near the path will often cause an AGV to stop.
Let’s say the area was clear when the AGV was installed, but the team later decided to store pallets right along the side of the path. This will cause the AGV to do one of two things – the first more benign than the second. The AGV will either slow down to a crawl while passing the obstacles, or drive by the obstacle at a high speed, without leaving room for a person to egress. The first scenario kills time. The second scenario is a major safety concern. By comparison, because an AMR is a path planning robot, it does not need to adhere to a fixed route and uses that ability to provide more space to obstacles near the path, with ample room for people to safely move about.
In addition, should an AMR run into a heavily congested area, it can navigate through tight spaces and make its way to the other side. What’s more, an AMR can sense dynamic obstacles, predict their future motions, and slow down to give way if needed.
Additionally, in most dynamic warehouse environments, goods are moving quickly, and workflow or floor plans change frequently. A path following AGV is not robust to these changes. A site manager must manually re-route the AGV to teach it a new route. These manual interventions are cumbersome and consume workers’ time.
Localization is the robot’s ability to know where it is in space, usually based on landmarks. Traditional AGVs, Advanced AGVs, and AMRs all localize in different ways, resulting in different levels of efficiency.
Natural Feature Localization
Traditional AGVs require reflectors or mirrors for localization, meaning additional infrastructure is required to install these self-driving forklifts. Advanced AGVs use what’s referred to as natural features for localization. Natural features are static things in the environment, like racking, walls, or pillars. With either Traditional AGVs or Advanced AGVs, both methods require the landmark to be visible and exactly the same. The reason is because AGVs need to know precisely where they are so they know they’re precisely on the right. If it cannot localize, the robot will stop and require manual intervention to get it started again. Once an AGV leaves its path, it has no way to get back on route.
Natural Feature Localization paired with Obstacle Avoidance
AMRs use natural feature localization, as well, but when paired with path planning as described above, they are much more robust to changes in their environment. When an AMR strays from the desired path, it is confident that it can use path planning and obstacle avoidance to move safely forward, avoid people and obstacles and get back on path. This approach saves not only travel time, but also the time that workers are required to help a lost AGV find its way.
The way Vecna Robotics has developed its localization is based on something similar to how a human localizes. The AMR just need to make progress down the aisle, but it doesn’t have to know precise measurements. Think about it. If you’re moving from you living room to the kitchen to get a glass of water, you don’t need to know in fractions of an inch how far you have to walk. All you need to know is roughly how to get there. Thus, the AMR moves from one area to another in a more natural way without requiring precise localization all the time.
Now, when you arrive in the kitchen, you do need to precisely where the glass of water is in order to pick it up and lift it to your lips. Similarly, when an AMR is ready to dock to a pallet or conveyor, it localizes more precisely.
The benefits to this method are that AMRs are more flexible to changes in their environment. An AGV is going to expect that a warehouse is always exactly the same as it was when it was first mapped. Because the environment is constantly in flux, performance over time will gradually become worse and worse. An AMR that relies on a combination of smart localization and obstacle avoidance will never degrade in performance – and in fact, has the potential to improve over time.
High-capacity AMRs and AGVs have one primary goal: move pallets efficiently. However, AMRs and AGVs are not created equally when it comes to pallet handling. This is a key area to understand when selecting a solution to maximize efficiency. Let’s get into it.
Independent Pallet Pick Up
One important distinction is how self-driving forklifts approach pallet pickup. Most traditional AGVs are unable to pick up pallets and require a human to manually load the pallet onto forks. Advanced AGVs are more independent and many can pick up pallets without manual intervention. However, AMRs are the most independent and consistently successful in this area. Needless to say, independent pallet pick up is critical to overall efficiency, so workers don’t have to stop what they’re doing and assist the AGV.
The other major factor that contributes to pallet handling efficiency is pallet detection. AGVs are designed to navigate to a location where they think their target pallet should be. If the pallet is placed askew, or if the pallet is not in that exact location, the AGV will still attempt the pick until it realizes nothing has been loaded onto its forks. It will then wait for help from a human worker. We all know that a warehouse environment is far from precise and perfect where goods and people are moving at lightning speed.
While AGVs use localization to find pallets, AMRs make use detection – that is, positive confirmation via onboard sensors - to determine if the pallet is there before they attempt the pickup. AMRs will travel to the area that they believe the pallet should be and look around for the right label just as a human would scan supermarket shelves for their favorite brand of soda. If the pallet is askew, or slightly out of the drop zone, the robot will be able to identify it, just as a person can see that a case of soda is at an angle on the supermarket shelf. If the right pallet is not where the AMR anticipated it should be, it will move to the next drop area and search again. This approach unlocks a massive amount of efficiency gains as it allows operators to work more quickly with less specificity around how and where they drop off pallets.
Using AMRs to Maximize Operator Efficiency
When comparing AMRs and AGVs, it’s important to take navigation into account. The gains created by path planning, obstacle avoidance, and pallet handling give AMRs the upper hand when it comes to efficiency and flexibility. Perhaps more importantly, AMRs allow operators the freedom to work without interruption with fewer stops and calls for local assistance, and the ability to forgive some imperfect pallet placement in pick and drop zones. AMRs are the key to maximizing operator efficiency.