• Pay: $54-$56
• Length of assignment: 12 month CTH
• Location: Warren, MI
• Visa: USC / GC

Ideal candidate will have a PhD in Robotics with experience working with embedded systems / Robotics Operating System (ROS). Most of the work is software development, mainly with C/C++. More systems experience required but for the machine learning part some Python experience would be helpful.

Hybrid 50/50 onsite at GM Tech Center in Warren, MI and remote.

The following thrust areas are defined as part of this work:

1. Development of Real-Time and offline data source credibility monitoring algorithms
2. Development data source selection solution/policy and performance evaluation
3. Development BEV (birds' eye view) representation for SLAM algorithms
4. Develop and evaluation place recognition solutions
5. Develop and evaluation the early stage fusion solutions

Consultant will leverage GM's network-based structures and real-time architecture to develop a comprehensive suite of embedded software solutions. These solutions will aim to address various challenges faced in autonomous driving, including:

• Data active insourcing policies: The supplier will explore different data active insourcing policies to strike a balance between performance and communication cost. This will ensure that the AVs receive accurate and up-to-date information about their surroundings while minimizing communication costs.
• Real-time data processing and analysis: The supplier will process and analyze the captured data in real-time, which is crucial for the effective functioning of autonomous vehicles.
• SLAM algorithms and crowdsourcing-based map building: The supplier will investigate and evaluate various simultaneous localization and mapping (SLAM) algorithms and crowdsourcing-based map building techniques. These techniques will help in constructing accurate 3D maps of the surroundings in real-time.
• Bird's Eye View representation: The supplier will assess the current state-of-the-art bird's eye view (BEV) representation and measure its performance in integrating data from different sensor types, such as mono cameras and radars.
• Perception and localization algorithms within the framework of a distributed networked sensor system: The objective of this exploration is to utilize the data from the networked sensors to improve the accuracy and reliability of the perception and localization algorithms. This will allow for the autonomous system to have a more robust understanding of its surroundings and thus, improve its ability to make informed decisions. In essence, the supplier will be exploring how connectivity can enhance the autonomy applications by leveraging the information from the networked sensors.