C

Caterpillar

Makes heavy machinery, engines and gas turbines

Autonomy Map Engineer

Computer VisionPosted on Nov 12
Full-timeNo salary listedIn-person

About the Job

Requirements

  • A Bachelor's degree or above in a relevant field such as surveying and mapping engineering, computer vision, remote sensing science, GIS, autonomous driving, or computer science; a Master's degree or above is preferred.
  • Three to six or more years of research and development experience in autonomous driving or high-definition mapping, including practical delivery experience in change detection, multi-source fusion, or high-definition map updates, and end-to-end algorithm development, validation, and production delivery.
  • Proficiency in C++ and Python.
  • Familiarity with the Linux development environment.
  • Proficiency with OpenCV, PCL, and Eigen for vision and point-cloud processing.
  • Familiarity with PyTorch or TensorFlow, including the ability to train and optimize image and point-cloud models.
  • Expertise in one or more of point-cloud processing and registration (ICP, NDT, feature matching), multi-sensor spatiotemporal fusion, 3D reconstruction and map generation, change detection, or autonomous-driving map updates.
  • Strong understanding of high-definition map data models and production workflows.
  • Familiarity with OpenDRIVE and Lanelet2 map standards and ability to extend standard map semantics for mining environments.
  • Familiarity with road-topology modeling, lane-network construction, and map version management.
  • Understanding of how localization and planning/control systems use maps and their precision requirements.
  • Ability to independently design technical solutions for automatic map updates and lead end-to-end R&D from requirements analysis and algorithm design through model training, system integration, and production release.

Responsibilities

  • Develop core algorithms for automatic updates of high-definition maps in open-pit mining areas, covering perception, change detection, multi-source fusion, and incremental map updates to support real-time, high-precision map requirements for autonomous mining-truck dispatch, localization, path planning, and motion control.
  • Develop map-change detection algorithms using vehicle perception data, LiDAR point clouds, visual images, and drone survey imagery.
  • Automatically identify and update map features including road boundaries, traffic infrastructure, road markings, loading and unloading points, and stopping areas.
  • Improve change-detection model robustness and generalization in dust-obscured scenes, around dynamic material piles and temporary construction areas, and in adverse weather.
  • Build a change-recognition system combining deep learning and geometric constraints to reduce false negatives and false positives.
  • Fuse crowdsourced vehicle perception, onboard LiDAR, cameras, drone surveys, remote-sensing imagery, and mining production data.
  • Establish a unified spatiotemporal reference framework for automatic alignment and fusion of heterogeneous map data.
  • Develop data association and consistency-optimization methods across multiple sensors and time scales.
  • Correct data errors caused by sensor noise, localization drift, and time synchronization issues.
  • Design and implement an incremental map-update framework, match old and new map features, analyze differences, and automatically merge updates.
  • Automatically repair road-topology relationships and maintain consistency between map semantics and topology.
  • Build high-definition map lifecycle management for dynamic mining environments.
  • Establish automated map-evaluation tools and metrics for update precision, recall, topological completeness, and geometric deviation.
  • Develop map-quality monitoring and regression-validation systems to support map-version release and production acceptance.
  • Develop algorithms from research prototypes through production systems; collaborate with perception, localization, planning and control, map-engine, and cloud-platform teams on system integration.
  • Deploy algorithms as services, optimize performance, and build capacity for large-scale data processing.
  • Increase automation to reduce manual review and map-maintenance costs.

Desired Qualifications

  • Familiarity with dynamic terrain characteristics and production workflows in open-pit mines.
  • Research and development experience in environments with low GNSS coverage and unstructured roads.
  • Experience delivering crowdsourced high-definition map-update projects.
  • Experience with map cloud platforms or large-scale map-data processing.
  • Experience deploying algorithm services, microservice architectures, and MLOps.
  • Experience reducing the proportion of manual map verification or building automated mapping platforms.

About the Company

Caterpillar

Makes heavy machinery, engines and gas turbines

Caterpillar makes construction and mining machines, diesel and natural gas engines, industrial gas turbines and locomotives for contractors, mines, energy companies and data center builders. It is the world's largest maker of construction and mining equipment, and its yellow Cat bulldozers and excavators work in nearly every country. Customers buy or rent machines through independent Cat dealers, while its engines and turbines generate electricity and push natural gas through pipelines. Caterpillar earns money from equipment sales and from the parts and repairs that dealers supply over a machine's working life. It was formed in 1925 when California's Holt and Best tractor companies merged, and it is based in Irving, Texas.

Company size
10,001+
Company stage
IPO
Headquarters
Irving, TX
Founded
1925

Recently Posted Jobs

Explore more opportunities at Caterpillar

See more jobs