Polynomial Trajectory Generation for Kinematic Smoothness

27 concepts22 papers1 proposal

When moving a physical system—whether it is a robotic arm, an autonomous vehicle, or an industrial crane—from point A to point B, the path cannot simply be a straight line or a sequence of sharp turns. In the physical world, forces dictate motion. An instantaneous change in velocity requires infinite acceleration, and an instantaneous change in acceleration requires infinite jerk (the third derivative of position). These mathematical discontinuities manifest as mechanical shock, violent vibrations, actuator saturation, and structural wear.

To achieve fluid, efficient, and safe motion, engineers must design trajectories that are continuously differentiable ($C^1$, $C^2$, or higher). Polynomial trajectory generation solves this problem by mapping the physical constraints of a system directly onto the algebraic properties of polynomial equations.

The Boundary Condition Match: Why Polynomials?

The core intuition behind using polynomials for trajectory planning is simple: the degree of a polynomial determines its degrees of freedom, which must match the number of boundary conditions required by the physical task.

If you only care about where a motion starts and ends, you have two constraints (initial and final positions). A first-degree polynomial (a straight line) satisfies this, but it forces an instantaneous jump in velocity at the boundaries.

Cubic Polynomials ($3^{\text{rd}}$-Degree)

To ensure a smooth start and stop, you must specify four boundary conditions: initial position, final position, initial velocity (usually zero), and final velocity (usually zero). A cubic polynomial provides exactly four controllable coefficients:

$$s(t) = at^3 + bt^2 + ct + d$$

In a foundational paper, Richard Paul demonstrated that employing third-degree polynomials ensures continuity in position, velocity, and acceleration ($C^2$ continuity) across path segments, eliminating sudden mechanical jerks. Early space exploration systems also relied on this property; Richard A. Lewis et al. utilized a sequence of five cubic polynomials for joint motion, noting that this formulation provides predictable movement while avoiding the extraneous "wandering" or oscillations characteristic of higher-degree polynomials.

Even in modern reinforcement learning frameworks, where policies learn complex whole-body behaviors, Dianyong Hou et al. use cubic polynomial interpolation to map reference trajectories. This simple mathematical constraint prevents the control policy from attempting instantaneous, destabilizing jumps to distant coordinates.

Quintic Polynomials ($5^{\text{th}}$-Degree)

While cubic polynomials smooth out velocity transitions, they still allow sudden jumps in acceleration at the boundaries. If a system must start and stop with zero acceleration to completely eliminate initial and final jerk, it introduces two more boundary conditions, totaling six. This requires a fifth-degree, or quintic, polynomial:

$$s(t) = at^5 + bt^4 + ct^3 + dt^2 + et + f$$

As Chao He et al. explain, a quintic polynomial trajectory planner provides the necessary degrees of freedom to satisfy these stringent boundary conditions, enabling precise control over position, velocity, and acceleration simultaneously.

Depending on the mechanical architecture, intermediate degrees can also be targeted. For instance, Jun Chi et al. utilized 4th-order polynomial interpolation for automated-cutoff valve cam profiles, demonstrating that four degrees of freedom can perfectly balance rapid response times with smooth mechanical transitions.

Why It Works: Analytical Efficiency and Minimum Jerk

Beyond matching boundary conditions, polynomials possess two unique traits that make them the industry standard for real-time trajectory generation: computational speed and an inherent relationship with physical optimization.

1. Closed-Form Analytical Solutions

Finding the coefficients ($a, b, c, \dots$) of a trajectory matching specific boundary conditions reduces to solving a system of linear equations ($Ax = b$). Because the matrix $A$ is derived purely from the chosen time intervals, this system can be solved analytically in closed form.

This eliminates the need for expensive, iterative numerical optimization loops. For example, Linzhu Yue et al. leveraged these analytical solutions to build a minimum-jerk trajectory planner for quadrupedal jumping that computes trajectories in just 50 microseconds, allowing the robot to adapt dynamically to obstacles mid-stride.

2. The Minimum-Jerk Principle

In physics and calculus of variations, it can be proven that the unique mathematical curve minimizing the total squared jerk ($\int \ddot{x}^2 dt$) over a given time interval—subject to fixed boundary positions, velocities, and accelerations—is precisely a quintic polynomial.

By using piecewise quintic polynomials, frameworks like the one proposed by Long Xu et al. ensure that trajectories are four times continuously differentiable. This minimizes curvature variations and jerk, allowing car-like robots to maintain stability and high execution speeds even when navigating complex, uneven terrain. Fundamentally, as shown by Xiaoyang Gu et al., engineering these piecewise polynomial segments acts as a direct lever to control acceleration and velocity, forming the bedrock of a curve's geometric and computational properties.

Tailoring Trajectories Across Domains

Because polynomials allow designers to explicitly shape velocity and acceleration profiles, the technique is widely adapted across diverse engineering challenges.

Autonomous Vehicles and Traffic Flow

For self-driving cars, passenger comfort and vehicle stability depend on avoiding sudden lateral forces. Liu X et al. utilize quintic functions to define trajectories for individual maneuvers like lane changes or merges, ensuring the overall path avoids abrupt changes in acceleration. To handle complex traffic environments, Rui Bai et al. advanced this by decomposing a single lane change into two sequential quintic polynomial segments, allowing granular control over separate phases of the maneuver.

On a macro scale, Sahand Khoshdel et al. employ high-degree polynomial functions to model aerial vehicle paths at intersections, ensuring that vehicles adhere to strict rhythmic timing constraints while minimizing energy-intensive accelerations.

Heavy Machinery and Precision Manufacturing

In heavy industry, smooth trajectories prevent catastrophic structural fatigue. Wensheng Zhang integrates quintic polynomials into the path planning of unmanned overhead cranes, enforcing boundary conditions that prevent heavy payloads from causing abrupt mechanical stress.

For precision tasks like robotic welding, Yun Shi et al. apply quintic curves to maintain continuous velocity and acceleration profiles, which is vital for weld quality. Furthermore, Weijun Wang et al. demonstrate that fifth-order polynomials can be shaped to strategically place velocity peaks to coincide with high-load regions, mimicking human biomechanics to reduce a construction robot's peak energy consumption by 48.4% without inducing mechanical wear.

High-Precision Fluidics and Aerospace

The utility of polynomial shaping extends to micro-scale and zero-gravity environments:

  • Liquid Handling: In high-throughput medical diagnostics, Jun Lu et al. utilize 5th-order polynomial acceleration profiles to control multi-axis pipetting assemblies. The resulting smooth transitions eliminate the micro-vibrations that cause specimen splashing and cross-contamination.
  • Space Robotics: For orbital assembly, the authors of Self-reconfiguration Strategies for Space-distributed Spacecraft refine manipulator movements using fifth-degree polynomial interpolation. This ensures that transitions between joint states calculated via kinematics remain continuous, avoiding structural stress on delicate modular components in space.

Key Foundational Work

Trajectory Control of a Computer Arm (1971) — contributed 2 concept elements to this direction.
Trajectory Control of a Computer Arm (1971) — contributed 2 concept elements to this direction.
Planning Considerations for a Roving Robot with Arm (1973) — contributed 2 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_polynomial_trajectory_generation_cd1448a3,
  title = {polynomial trajectory generation},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/polynomial-trajectory-generation/rdr_cd1448a38f354023b37517d983bac818},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-21},
}

This exact version (version 1, frozen text):

@misc{lacuna_polynomial_trajectory_generation_cd1448a3_v1,
  title = {polynomial trajectory generation},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/polynomial-trajectory-generation/rdr_cd1448a38f354023b37517d983bac818/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-21},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake