This is Patent Pending. Supply chain-related patent applications are published every day and this is where we'll talk about the ones that could have the biggest impact on the supply chain and the ones that challenge the norm. We want to give you an idea of where supply chains are heading and what the industry is thinking. Read the last issue here.
Route the order right
There are a variety of reasons why a company would want to optimize delivery routes to customers: the price of fuel, speed, labor cost. Walmart envisions a system to reduce costs using historical data on inbound and outbound deliveries to optimize load and route planning, according to a patent application from the retailer published this month.
But the system is not just about finding the shortest route: It's about reducing the number of empty miles delivery drivers travel or drive without an order. The system is able to assign new orders to vehicles that might already be on the road making a delivery. It can reduce empty miles by helping ensure a vehicle is carrying an order more often than not, so instead of sending drivers back to where they started, it would send them to the location of the new order. Walmart does not say how much it would expect empty miles to be reduced using this technique.
Other companies are testing similar processes. Convoy said its Automated Reloads system, which does something similar, resulted in a 45% reduction in carbon emissions by reducing the number of empty miles truckers travel.
Read up:
- Convoy: Bundling freight cuts carbon emissions 45%
-
76% of shippers using route, load optimization in sustainability efforts
Get a grip, robots
Warehouse management system, order management system ... what about a grip management system? This is what Amazon proposed in a patent application published this month. Robots would use the set of information to know what grip to use when trying to grasp objects.
Robots, Amazon explains, struggle with inconsistency, but are great at repetitive tasks.
There are a number of options a robot can consider when picking up an item. Where should it grip? How should it orient the item while carrying it? How hard should it grip? What gripping function should it use? If it gets any of criteria wrong, it could drop or break the line item.
Amazon uses the example of a teddy bear. A grip management system would store the information taught to the robot about how to pick up a teddy bear. If it was taught to pick it up by the leg, the system could then extrapolate based on that information that it could pick the teddy bear up by other cylindrical objects — like the arms. The knowledge of cylindrical objects to lift items could extend beyond teddy bears to other, similar items with the grip management system.
The idea would be to help the robot adjust its own grip for new objects it's never seen before. So when it sees a new item it would determine the right grip by calculating the probability of success based on what it's been taught. Sensors on the robotic arm would provide the system with feedback on the grip and contact point used, allowing it to update its probabilities of success when it sees that item again in the future.
Read up:
Cluster picks
What's the fastest way to pick items? It's something a lot of managers have spent time trying to figure out and it's not the first time I've written about it in this column. But what Locus Robotics proposes in its patent application published last month is a little bit different. Instead of finding the shortest path for a picker to get to an item, it uses cluster analysis to group items in orders.
Locus would cluster items within orders as they arrive in the warehouse management system. The clusters are based on the item's location in the store, the system determines an item's cluster with its X and Y coordinates within a grid covering the entire floor space. (The closer items are to each other within the warehouse, the more likely it is they'll be in the same cluster for picking.) Item groups are then assigned to either a human (via a handheld device) or a robotic picker (via an onboard electronic device). The assignments are based on the carrying capacity of the robot or the worker.
