ADAS Perception Engineer

A global semiconductor and embedded-technology organisation is seeking an ADAS Perception Engineer to develop a production-focused, camera-based lane detection and departure warning solution.

The organisation develops advanced technologies for automotive, industrial, Io T and connected applications, combining embedded processing, edge AI, security and mixed-signal expertise.

You will contribute to technology designed to make vehicles safer and more intelligent.

This assignment covers the complete perception pipeline—from image pre-processing and neural-network inference through temporal tracking, lane geometry estimation and Lane Departure Warning logic.

• What You’ll Be Doing
• Translate functional requirements into a front-camera lane detection architecture, defining inputs, outputs, latency, accuracy and operational design domain targets.
• Design, train and evaluate deep-learning models for detecting lane markings and boundaries.
• Explore segmentation, anchor-based, row-classification, polynomial and transformer-based approaches.
• Implement temporal lane tracking using techniques such as Kalman filtering, extended Kalman filtering, particle filtering or learned temporal models.
• Improve perception stability in challenging scenarios, including occlusions, faded markings, missing lanes, construction areas, sharp curves, rain, glare and low-light conditions.
• Fit and maintain polynomial or clothoid-based lane models.
• Estimate the vehicle’s lateral position and heading relative to detected lanes.
• Develop Lane Departure Warning functionality, including time-to-lane-crossing calculations, warning thresholds, driver-intent filtering and turn-signal suppression.
• Integrate intrinsic and extrinsic camera calibration, homography and inverse perspective mapping.
• Optimise models for real-time deployment on automotive embedded hardware using techniques such as quantisation, pruning and accelerated inference.
• Define annotation guidelines, dataset requirements and data-collection strategies.
• Establish validation KPIs covering detection range, curvature accuracy and false-positive and false-negative performance.
• Support Euro NCAP-aligned LDW/LKA testing and collaborate with integration teams during HIL and vehicle-level validation.
• Your Background
• Strong professional experience in computer vision and deep learning for automotive perception.
• Practical knowledge of semantic segmentation, keypoint detection or curve-based lane detection.
• Proficiency in Python and Py Torch.
• Strong C++ skills for production or embedded implementation.
• Experience developing lane or object tracking systems using Kalman filters, EKFs, particle filters or comparable temporal-fusion methods.
• Knowledge of camera calibration, homography, inverse perspective mapping and classical computer-vision techniques.
• Experience with polynomial or clothoid-based lane geometry modelling.
• Experience deploying and optimising neural networks for embedded or automotive compute platforms.
• Understanding of ADAS software architecture and real-time system constraints.
• Familiarity with Euro NCAP LDW/LKA protocols and awareness of Automotive SPICE processes.
• Experience working with datasets such as Tu Simple, CULane, BDD100K or proprietary automotive datasets.

By applying to this role you understand that we may collect your personal data and store and process it on our systems.

For more information please see our Privacy Notice (https://eu-recruit. com/about-us/privacy-notice/)

In accordance with local employment laws, applicants must have current, valid authorisation to work in Germany at the time of application.

We are unable to sponsor employment visas for this role.

Applications from individuals without existing work authorisation for Germany cannot be considered.

#J-18808-Ljbffr