High-Precision Lane Markings at
Scale for Autonomous Trucks
Boosting autonomous truck safety with precise, scalable lane marking powered by expert annotators and custom tooling.
Challenge
In a world where autonomous vehicles are advancing faster than most expected, accurate lane detection is critical, especially for the heavy-duty autonomous trucks that are set to revolutionize the logistics industry. When a leading trucking company, one of the pioneers in long-haul autonomous transport, partnered with iMerit, they brought challenges that pushed beyond industry norms. Each scene consisted of 350-frame sequences and over 75 million LiDAR points, a massive dataset that required annotations extending far beyond typical horizon distances.
Their ambitious goal was to achieve 200,000 lane markings per week without a proportional increase in manpower. This required iMerit to rethink speed, scale, and precision from the ground up to empower a new generation of safer, more efficient autonomous trucks.
“iMerit’s incredibly knowledgeable data labelers are part of our team and are the right people for our enrichment efforts.”
Solution
Given the scale and complexity of the data, human effort was strategically augmented with innovative tools:
- Merge Point Cloud (MPC) Tool: Allowed for seamless long-sequence annotations.
- Smart Validations: Auto-flagged geometry errors in real time.
- Ground Validation & Auto-Grounding: Ensured accurate road-surface alignment.
- Intensity Control Sliders: Made faint lane markers visible in noisy point clouds.
- Backend Optimization: Enabled smooth navigation through massive datasets.
To meet the client’s aggressive delivery and quality benchmarks, iMerit initiated a comprehensive plan that combined specialized human expertise with cutting-edge automation.
Our team created tailored training focused on real-world lane marking complexity, including geographic diversity, weather conditions, construction zones, and merging/splitting lanes. This upskilling effort drastically improved first-pass accuracy and minimized rework.
These software features helped maintain high annotation quality while significantly reducing time per task, creating a scalable solution for the client’s long-term vision.
Result
By pairing specialist talent with robust and AI-enhanced software, iMerit consistently delivered 200,000 lane annotations per week, aligning perfectly with the client’s ambitious quarterly targets. This allowed the client to hit critical development milestones and accelerate their time to market. The blend of specialized human expertise and advanced automation tools resulted in a significant reduction in the average annotation time, decreasing it by over 68%.
This not only accelerated data delivery but also brought about considerable cost efficiencies. By maintaining high consistency in lane labeling across long-range, dense sequences, iMerit empowered the client to develop highly robust perception models, a foundational element for the safety and reliability of their autonomous trucking advancements.
