Pallet detection system with a real-world dataset

Increased efficiency of robot fleets

Inefficient processing and inaccurate picks diminish robot fleet ROI.

Automated forklifts must navigate busy floors, handle imperfect pallets, and verify spaces—all while managing real-world complications like dirt, wear, and shrink wrap.

Pallet detection can accelerate missions by reducing complexity on both sides of moving a pallet. It efficiently localizes racking systems while handling obstacle detection and space verification.

Certainty in unpredictable environments 

Every deployment presents unique challenges and anomalies.

A pallet detection system with high accuracy and reliability overcomes the biggest challenge in pallet detection: the variability of pallet materials, colors, and conditions.

It accounts for countless nuances, unexpected problems, outlier geometries, and other variables, even across locations within the same company. A robust system provides close positioning and orientation precision for seamless pallet interactions.

Increasing ROI 

After years of innovation, the current BOM costs of an industrial mobile robot cannot decrease much further. But the price of a robot, let alone a fleet, is still cost-prohibitive for many companies. 

Robotics companies can deliver greater value through improved efficiency than through hardware cost reduction. Increased throughput – accomplishing more work with fewer units – justifies the capital expense of a mobile robot fleet for a facility. 

Faster fleets require fewer units, making them accessible to more facilities. End users benefit from increased production with minimal staffing, while robotics companies increase sales with a more valuable product.  

Read more: ROI calculations for mobile robotics.

Creating a fast, accurate, and reliable pallet detection system is a much harder task than most developers realize. ifm's seven years of development and innovation revealed three key complexities:

1. Complex variables: Success requires accounting for an abundance of factors beyond basic pallet characteristics.

2. Performance impact: Even delays as minor as one extra second per pick can prevent meeting daily operational targets.

3. Ongoing development: Systems need continuous resources to handle new outliers, edge cases, and pallet types.

Read more: Pallet detection challenges and overcoming them.

Make or buy: Evaluating the options

Creating a viable pallet detection system is extremely difficult though possible. The real question is: Does a proprietary system align with a company's strategic intellectual property (IP) goals? 

Despite advances in machine learning and synthetic data, development requires significant time and resources. Third-party solutions like ifm's Pick and Drop System offer immediate resolution. 

The latest generation of the ifm PDS solution delivers proven reliability with 10x faster performance, drawing from extensive real-world testing. 

An optimized, ready-to-integrate solution reduces development costs and frees developer capacity to focus on core IP development.

Read more: Making vs. buying pallet detection.

Increasing efficiency

The ifm Pick & Drop System improves autonomous vehicle performance. It reduces pallet handling time from 2+ seconds to 0.2 seconds while maintaining detection quality. Automated forklifts reliably select and place pallets in even the most demanding conditions.

Providing certainty 

The ifm algorithm accounts for countless nuances, unexpected problems, outlier geometries, and other variables. They provide 1-centimeter precision in positioning and orientation, ensuring seamless pallet interactions.

The above pallets are examples of those detected by PDS, but it is not limited to these pallet types.

Pick & Drop System hardware upgrades

ifm continues to advance and adapt. The newest generation of the Pick & Drop System now runs on the O3R robotic perception platform. This new system architecture increases computational power by moving the algorithm from onboard the camera to a separate processing platform.

Powered by NVIDIA Jetson platform® it delivers faster picks through enhanced computational capability and greater flexibility.

The benefits of real-world experience 

Built on millions of picks, the ifm PDS software transforms complex, real-world pallet geometry data into actionable intelligence. The continuously evolving algorithm adapts to new challenges through regular updates, delivering unique value to autonomous vehicle operations.

ifm provides comprehensive support through the PDS vision assistant software. When encountering unique scenarios, customers can record 4-5 frames of new operations for analysis. ifm helps optimize parameters and integrate custom pallet types, continuously improving system performance.

Get started with the ifm Pick & Drop System.

Pallet detection requires datasets to recognize different pallet types and conditions. They operate using one of two models: 

• A heuristic method based on “real data” drawn from real-world scenarios and dynamics

• AI machine learning using synthetic data as training data. 

Heuristic methods employ defined rules and direct geometry detection, offering transparency and consistency. While this approach eliminates retraining cycles, it demands extensive real-world data collection. 

Using the heuristic approach, ifm developers quickly update and improve performance without the guesswork of a "black box" algorthmic framework. The ifm Pick and Drop System regularly incorporates new information, with nearly a decade of data from over 10 million successful picks across 1,700+ vehicle installations.

The future of pallet detection

Mobile robots have yet to cross the chasm into mass adoption. However, robust features like pallet detection and obstacle detection quickly increase the efficacy and value of AMRs. 

As pallet detection systems improve, robots will be able to work faster and with more accuracy. This will increase their value, making them more affordable to both large- and mid-size companies.

While each company may require fewer robots in a fleet to complete their work, this significantly increases the number of potential customers and helps drive increased demand overall.