What Is Trajectory Tracking?

Trajectory tracking is the control problem of steering a dynamical system so that its state follows a prescribed time-parameterized reference trajectory as closely as possible despite disturbances, model uncertainty, and actuator limitations. The reference trajectory specifies desired position, velocity, and sometimes acceleration as explicit functions of time, and the tracking controller must generate corrective commands at each time step to minimize the deviation between the actual and desired state. Unlike point stabilization, which drives a system to a fixed equilibrium, trajectory tracking requires the controller to handle a moving target whose desired state changes continuously throughout the maneuver.

The problem is central to robotics, aerospace, and autonomous systems. In robotic manipulation, trajectory tracking determines whether a tool follows a welding seam or assembly path with acceptable precision. In autonomous vehicles, it governs how accurately the vehicle adheres to a planned lane or speed profile. In aerospace, it defines how well a guided vehicle follows an intercept or approach path. The theoretical foundations draw on Lyapunov stability theory, nonlinear control, and state estimation.

Tracking Error and Control Objectives

Tracking error is defined as the difference between the actual system state and the reference trajectory at each instant. A trajectory tracking controller aims to drive this error to zero, or to keep it bounded within an acceptable tolerance, while satisfying constraints on actuator inputs. Two standard performance objectives are asymptotic tracking, in which the error converges to zero as time progresses, and practical tracking, in which the error converges to a small neighborhood of zero bounded by a function of the disturbance magnitude. The tracking bandwidth, or how quickly the controller can respond to changes in the reference trajectory, is constrained by actuator dynamics and the sampling rate of the feedback loop. A review of trajectory tracking control for flexible-joint robots using Extended Kalman Filter and PD control demonstrates how state observers are needed when direct measurement of all relevant states is impractical, using filter-estimated velocities in place of numerically differentiated position signals.

Feedback Control Architectures

Classical trajectory tracking controllers add a feedforward term, derived from the reference trajectory dynamics, to a feedback error-correction term. The feedforward contribution compensates for predictable demands imposed by the trajectory itself, such as the torques required to accelerate a robot link, while the feedback term corrects for disturbances and modeling errors. Proportional-derivative (PD) and proportional-integral-derivative (PID) controllers are the most widely used feedback components in practice; their tuning for trajectory tracking involves selecting gains that provide sufficient damping without exciting resonant modes. In robot control, computed torque control augments PD feedback with a model-based term that cancels the nonlinear dynamics of the manipulator, reducing the tracking problem to one of linear error regulation. The ScienceDirect paper on combined trajectory tracking and path following for wheeled mobile robots examines how these two control objectives are unified into a single framework for underactuated vehicles.

Advanced Tracking Methods

For systems with significant uncertainty or operating at the limits of performance, advanced tracking methods extend beyond classical feedback. Sliding mode control drives the system state onto a sliding surface defined in error coordinates, then constrains the dynamics to that surface using high-frequency switching, yielding robustness to bounded disturbances at the cost of potential chattering in the control signal. Adaptive controllers update their parameters online in response to estimated parameter errors, useful when payloads vary or actuator characteristics drift. Model predictive control (MPC) solves a finite-horizon trajectory tracking problem at each sample step, allowing explicit constraint enforcement on actuator saturation and state bounds. A PLOS ONE study on tracking control of robotic manipulator end-effectors using robust sliding mode methods reports tracking error reductions of over 40 percent compared to conventional PD control in manipulators subject to payload variation and joint friction.

Applications

Trajectory tracking has applications in a wide range of fields, including:

• Industrial robot arm control for manufacturing and assembly
• Autonomous vehicle lateral and longitudinal control
• Unmanned aerial vehicle attitude and position control
• Surgical robot guidance in minimally invasive procedures
• CNC machine tool contouring accuracy
• Marine vessel dynamic positioning and course keeping