<![CDATA[Intercepting a highly maneuvering moving target remains a difficult planning problem because the target may change heading, accelerate, halt, or reverse without warning, while the pursuer only receives noisy position observations. This study proposes the Adaptive Pursuit Route Algorithm (APRA), an interception path-planning method that combines short-horizon target-state estimation, time-to-go aim-point projection, dynamically weighted cost terms, and a bounded residual that is learned online from past prediction errors. APRA is evaluated against three classical pursuit laws-Pure Pursuit, Proportional Navigation (PN), and Augmented PN (APN)-on a physics-based synthetic dataset of 3,200 trajectories spanning eight motion regimes, yielding 12,800 engagements. APRA attains the best result on every metric: an interception rate of 90.91% (95% CI 0.899-0.919), the shortest mean interception time (9.65 s), the lowest mean prediction error (8.73 units), the highest path efficiency (0.798), and the lowest control energy (48,022). One-way ANOVA confirms that the differences in time, energy, and prediction error are statistically significant (p < 0.001), and a Welch t-test shows APRA is significantly faster than APN (t = -14.48, p < 0.001). The results indicate that disciplined filtering combined with bounded online adaptation is more robust against deceptive and abrupt maneuvers than raw lead-based guidance.]]>
Copyrights © 2026
