Katkat
Back

Optimize Autonomy Mapping Components

ZiplineSoftware Engineer · Onsite
MediumSoftware EngineerOnsiteSoftware Engineering Fundamentals

Last updated

This question evaluates a candidate's skills in embedded perception and state-estimation, covering design of data structures and update algorithms for a 2.5D elevation grid with uncertainty as well as conceptual review of a causal one-dimensional smoothing filter.

An autonomy team runs perception and state-estimation code on a resource-constrained embedded computer. Discuss both of the following technical review tasks:

  1. Embedded elevation mapping optimization: The system receives timestamped 3D sensor points and vehicle poses. It must maintain a local 2.5D elevation grid containing estimated ground height and uncertainty for each cell. Describe the data structures, update algorithm, and embedded optimizations you would use to meet real-time CPU, memory, and latency constraints.

  2. One-dimensional filter review: The system also contains a causal one-dimensional filter for smoothing a noisy sensor signal. Review the implementation conceptually. What correctness, robustness, and testing issues would you look for, and how would you fix them?