Motion Planning Engineer Job Description

A Motion Planning Engineer builds the part of an autonomous system that decides how it moves. Give a robotaxi a destination, a warehouse robot a shelf to reach, or a robot arm a part to pick, and something has to choose the next maneuver, generate a path and speed profile, check it against every obstacle and agent nearby, and hand a clean trajectory to the controllers many times a second. That something is the planner, and this engineer writes, tunes and defends it.

The work splits into layers. Behavior planning (sometimes called decision making) picks the high-level action: yield or go at an unprotected left, change lanes now or wait, approach a shelf from the left or the right. Motion or trajectory planning turns that choice into a time-parameterized path that respects physics: curvature and acceleration limits for a vehicle, joint and torque limits for an arm, footstep constraints for a legged robot. Underneath sit search and optimization methods, from A* and lattice planners to sampling-based planners such as RRT and model predictive control formulations solved with numerical optimizers.

Aurora gives a sense of how central the function is inside an autonomy company. In a 2022 profile, a tech lead on its Motion Planning team said the team plays a key role in defining how the Aurora Driver safely navigates through the world, and that his own remit covered highway and surface street capabilities. In robotics, O*NET's task list for robotics engineers includes planning mobile robot paths and teaching path plans to robots.

A planner is only as good as its picture of what everyone else will do, so the job sits close to prediction. Waymo researchers describe reliable forecasting of road agents' future behavior as a critical component to safe planning in autonomous vehicles. Day to day, that means the motion planning engineer consumes predicted trajectories with their uncertainty, decides how cautious to be around each agent, and argues with the prediction team when forecasts and plans interact badly.

A typical week mixes four kinds of work:

• Algorithm and code. Writing and optimizing C++ that runs on the vehicle or robot under a hard latency budget, plus Python for analysis and tooling.
• Simulation and evaluation. Building scenarios, replaying logs, and reading metric dashboards for collisions, near misses, hard braking, jerk and progress.
• Incident triage. Pulling a log where the system hesitated or braked hard, reconstructing what the planner believed, and deciding whether the fix belongs in planning, prediction or perception.
• Safety evidence. Writing up design rationale and test results that feed the company's safety case before a release goes to the fleet.

How this differs from neighboring jobs matters when reading a posting. A perception engineer turns raw sensor data into objects and maps. A controls engineer makes the actuators track a commanded trajectory. A machine learning generalist on an autonomy team may train models anywhere in the stack. The motion planning engineer owns the decision layer in between, and is judged on whether the machine's motion is safe, legal, smooth and predictable to the humans around it.

O*NET places robotics engineers in Job Zone Four, where most occupations require a four-year bachelor's degree. As a rough guide, motion planning engineers come from mechanical engineering, electrical engineering, computer science, aerospace engineering or a dedicated robotics program. A master's or PhD focused on planning, optimal control or robot learning helps on research-heavy planning teams, while a bachelor's degree plus shipped planning code can carry a candidate on product teams.

The core technical foundation is mathematical:

• Linear algebra, probability and numerical optimization, including convex and nonlinear optimization, since trajectory generation is often posed as an optimization problem with constraints.
• Kinematics and dynamics, from bicycle models for road vehicles to forward and inverse kinematics for manipulators.
• Search and sampling algorithms, such as A*, D* variants, RRT and PRM, plus collision checking geometry.
• Control theory, especially model predictive control, because the boundary between planning and control is blurry in practice.

On the software side, the job is production engineering, not just research. O*NET's technology list for robotics engineers names C++, Python, MATLAB, Simulink, the Gazebo simulator, Linux and real-time operating system software. Open-source tools in this space include ROS 2 and its planning libraries MoveIt and Nav2, plus OMPL for sampling-based planning, and autonomy companies run their own simulators and log-replay tools. Comfort with profiling code, reasoning about memory allocation and threading, and writing deterministic tests is part of the bar.

Safety literacy is the other half of the background. UL 4600, whose third edition (Ed. 3-2023) was published March 17, 2023, covers how to build and evaluate a safety argument for vehicles that can operate in an autonomous mode, and planning engineers contribute evidence to those cases. On the industrial side, ANSI/A3 R15.06-2025 sets safety requirements for industrial robots and robot systems; it is a revision of ANSI/RIA R15.06-2012 and the US national adoption of ISO 10218-1:2025 and ISO 10218-2:2025. Neither is a personal credential, but planners that run near people have to be designed with them in mind.

There is no license or certification that defines this job. As a rough guide, routes in include graduate research, internships on autonomy or robotics teams, and student competitions such as autonomous racing and robotics challenges, with senior scope typically coming after roughly 5 to 8 years of shipping planning code. Useful signals in a portfolio include an open-source planner, a published paper with code, or a documented project where a robot or vehicle ran a planner you wrote on real hardware.

BLS does not count motion planning engineers separately, so the best federal view is the robotics engineer profile. O*NET projects average growth (3% to 4%) for robotics engineers from 2025 to 2035, with about 8,800 openings a year on average. Those figures describe a wide group of robotics engineers, not planning specialists alone.

Commercial deployment gives a more specific picture. Aurora reported in its third quarter 2026 shareholder letter that the Aurora Driver surpassed 500,000 commercial driverless miles in September 2026. Every one of those miles depends on a motion planner choosing lane position, speed and gaps in traffic without a human driver, and each new route, weather condition or vehicle platform brings new scenarios, edge cases and validation work for the planning team.

The robotics side shows activity too. North American companies ordered 10,989 robots in the third quarter of 2026, valued at $723.5 million, according to the Association for Advancing Automation (A3), and orders increased 24.8% in units compared with the third quarter of 2025. Robots that repeat fixed, taught paths need little online planning; mobile robots, flexible pick-and-place cells and humanoid platforms depend on it.

Regulation is moving in a direction that matters to planning teams. NHTSA has proposed changes to several Federal Motor Vehicle Safety Standards so they fit ADS-equipped vehicles, including those without manually operated driving controls. Proposed rules covered the windshield defrosting and wiping standards (FMVSS No. 103 and 104) on March 16, 2026, the light vehicle brake standard (FMVSS No. 135) on June 26, 2026, and the lamps and lighting standard (FMVSS No. 108) on Oct. 26, 2026. These were proposals, not final rules, when published. Vehicles designed without steering wheels or pedals put the full driving decision on the planner.

Because the title lives inside autonomy and robotics programs rather than across the whole economy, deployment and funding news from those programs is worth watching. The core skills of search, optimization and collision checking transfer between road vehicles, warehouse robots and defense ground vehicles, so a planning engineer can follow the work across markets.

Career paths split two ways. On the technical track, engineers move from owning a planner component to leading a planning area (highway driving, parking, manipulation grasp planning) and then to staff or principal roles setting planning architecture. On the management track, planning leads run teams that span behavior, trajectory and evaluation. Some move sideways into simulation and validation leadership, systems safety, or robot learning research, where the same skills apply.

Dear Hiring Manager,

I am applying for the Motion Planning Engineer position on your mobile robotics team. For the past three years I have worked on the planning stack for autonomous yard tractors at a logistics automation company, and I would like to bring that experience to a team building planners for robots that share space with people.

In my current role I own the trajectory optimizer that runs on our tractors. When I joined, the planner produced jerky speed profiles near trailers and docks, which drew complaints from site operators and caused nuisance stops. I reformulated the speed planning step as a quadratic program with explicit jerk limits, rewrote the hot path in C++ to remove per-cycle memory allocation, and built a regression suite of logged dock approaches in our simulator. The new planner shipped to every site after two months of shadow testing, and operator complaints about abrupt stops dropped noticeably.

I also spend a large part of my week in incident triage. When a tractor hesitates at a crossing, I pull the log, replay it, and work out whether the planner, the prediction model or the map caused it. That habit has made me careful about writing down why a planner behaves the way it does, which has helped our safety team assemble evidence for each release.

Before this job I completed a master's degree in robotics, where my thesis compared sampling-based and optimization-based planners for a seven-degree-of-freedom arm in cluttered bins. I am comfortable in ROS 2, MoveIt and OMPL, and I enjoy the moment when a planner that looks good in simulation finally behaves well on real hardware.

Your team's work on robots operating near warehouse staff is exactly the problem I want to spend the next several years on. I would welcome the chance to talk about how I could contribute.

Sincerely, Jordan Avery