Trajectory Planning with Simplified Vehicle Dynamics Models

40 concepts42 papers1 proposal

When designing motion planners for autonomous vehicles, robots, or unmanned aerial systems, engineers face a fundamental tension between physical realism and computational speed. To safely navigate a complex environment, a vehicle must calculate its future path within milliseconds. However, real vehicles are governed by highly non-linear, high-fidelity physics involving tire-road friction, chassis momentum, suspension forces, and powertrain lag.

To achieve real-time execution, researchers frequently rely on simplified vehicle dynamics models. By abstracting a complex physical system into a lower-dimensional mathematical representation—such as a point-mass, a first-order integrator, or a basic kinematic model—planners can transform intractable optimization problems into smooth, solvable equations. Yet, this computational efficiency comes at a cost: if the model is too simple, the planned trajectory may violate the physical limits of the actual vehicle, leading to control failure or collisions.


The Computational Appeal of Simplification

Why do planners deliberately ignore the rich physics of real vehicles? The primary reason is mathematical tractability. High-fidelity models introduce non-convex constraints and tightly coupled variables that slow down optimization algorithms.

By restricting control inputs or simplifying the underlying physics, researchers can decouple these complex systems. For instance, Tamás Kalmár-Nagy et al. demonstrated in Near-optimal dynamic trajectory generation and control of an omnidirectional vehicle that simplifying motor models and assuming no-slip conditions allows a complex non-linear trajectory planning problem to be broken down into a simpler, decoupled form suitable for real-time applications.

Similarly, reducing a vehicle's motion to first-order dynamics allows researchers to apply elegant optimization frameworks from other fields. In their work on intersection safety, Heejin Ahn et al. mapped the collision avoidance problem to a job-shop scheduling task in Semi-autonomous Intersection Collision Avoidance through Job-shop Scheduling. This formulation relies on first-order dynamics to ensure the supervisor can compute safety overrides in real time, though it introduces a distinct gap between the model and physical reality.

[High-Fidelity Physics] --------(Abstraction)--------> [Simplified Models]
(Tire slip, chassis roll,                             (Point-mass, kinematic,
 engine lag, fluid drag)                               first-order ODEs)
          |                                                    |
          v                                                    v
  Intractable Online                                   Real-Time Feasible
  Optimization (Slow)                                  Optimization (Fast)

Where Abstractions Break Down

While simplified models enable real-time planning, they introduce systematic vulnerabilities by ignoring critical physical constraints:

This challenge extends beyond ground vehicles. In underwater and aerial domains, modeling agents as simple points or single integrators fails to capture fluid-structure interactions, heavy currents, or aerodynamic flight characteristics, as noted respectively by Xiaomin Lin et al. in OysterSim: Underwater Simulation for Enhancing Oyster Reef Monitoring and Gabriel Barsi Haberfeld et al. in Risk-Sensitive Rendezvous Algorithm for Heterogeneous Agents in Urban Environments.


Bridging the Fidelity Gap

To leverage the speed of simplified models without sacrificing safety, the research community employs several bridging strategies:

1. Upgrading to Non-Holonomic Kinematic Models

Instead of treating a vehicle as a freely moving particle, planners can incorporate non-holonomic constraints that dictate how a vehicle can actually steer. Hassan Jafarzadeh et al. suggest in Learning Model Predictive Control for Connected Autonomous Vehicles that moving from a point-mass model to Dubins path dynamics—which enforce a minimum turning radius—significantly improves the real-world applicability of control loops.

2. Bi-Level Optimization and Data-Driven Refinement

High-performance scenarios like autonomous racing require a tight coupling between speed and safety. In FSDP: Fast and Safe Data-Driven Overtaking Trajectory Planning for Head-to-Head Autonomous Racing Competitions, Cheng Hu et al. utilize a kinematic model within a bi-level quadratic programming approach to maintain real-time execution, while acknowledging that layering more accurate dynamic models over the optimization framework is the key to unlocking higher performance limits.

3. Co-Simulation Frameworks

Another prominent trend is the use of modular testing pipelines. Rather than rewriting the planning algorithm to handle complex physics, researchers deploy the simplified planner within a co-simulation environment. As detailed by Yi Guo et al. in OpenCDA: An Open Cooperative Driving Automation Framework Integrated with Co-Simulation, a basic vehicle dynamics model can handle the high-level cooperative planning, while specialized simulators like CARSim validate the resulting maneuvers against high-fidelity stability and control constraints before real-world deployment.

Key Foundational Work

Autonomous Exploration In Unknown Urban Environments For Unmanned Aerial Vehicles (2005) — contributed 1 concept elements to this direction.
Near-optimal dynamic trajectory generation and control of an omnidirectional vehicle (2004) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_simplified_vehicle_dynamics_6375b7b3,
  title = {simplified vehicle dynamics},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/simplified-vehicle-dynamics/rdr_6375b7b3095b4a18ab40c5f87d225ebf},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_simplified_vehicle_dynamics_6375b7b3_v1,
  title = {simplified vehicle dynamics},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/simplified-vehicle-dynamics/rdr_6375b7b3095b4a18ab40c5f87d225ebf/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake