Driving · 2026-11-13

APG-Planner: Learning Vision-Based High-Speed Trajectory Planning via Analytic Policy Gradient

Jinwoo Lee, Geunsik Bae, Changseung Kim, Jiwoo Kim, Woojae Shin, Hyondong Oh

arXiv:2611.15296PDF

AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.

TL;DR

APG-Planner is a vision-based high-speed trajectory planner trained with analytic policy gradients (APG). The method uses differential flatness to analytically reconstruct vehicle states from planned trajectories, enabling differentiable trajectory-level state transitions without explicitly modeling low-level dynamics. The resulting differentiable episode rollout supports backpropagation through the planning sequence and avoids noisy sample-based policy-gradient estimation. After ~30 minutes of training on a single GPU, it reaches up to 11 m/s in real flight.

Why it matters

The authors address noisy, sample-intensive training of model-free policy gradients for high-speed vision-based planning. By combining differential-flatness-based analytic state transitions with APG and BPTT, the planner can optimize episode-level losses over successive receding-horizon replanning decisions without an explicit low-level dynamics model, and it aims for interpretable planned trajectories usable with existing low-level controllers.

Method

  • Train a trajectory-planning policy that outputs terminal conditions for world-frame trajectory segments (position via quintic polynomial; heading from horizontal tangent).
  • Construct differentiable receding-horizon state transitions using differential flatness: evaluate the planned trajectory at ΔT and reconstruct the next planning state via the flatness map (no explicit low-level dynamics unrolling).
  • Compute policy gradients analytically via APG by building a differentiable episode rollout and backpropagating episode-level losses with BPTT through the proposed state transition model (with temporal-gradient stabilizations).

Limitation

The authors train using differential-flatness-based state transitions that assume accurate trajectory tracking.

Abstract (from arXiv)

Vision-based high-speed navigation requires a trajectory planner that can reactively generate smooth, collision-free trajectories from onboard visual observations. Deep reinforcement learning offers a promising approach to vision-based trajectory planning, as it enables the optimization of sequential decision-making policies in complex environments. However, model-free policy gradient methods rely on sample-based policy gradient estimation, which yields noisy gradient estimates and inefficient training. To this end, we propose a framework for learning trajectory planning policies using the analytic policy gradient (APG) method and use it to train APG-Planner, a vision-based high-speed trajectory planner. The proposed framework leverages differential flatness to analytically reconstruct vehicle states from planned trajectories, thereby enabling differentiable trajectory-level state transitions without requiring an explicit low-level dynamics model. By iteratively performing trajectory planning and differentiable state transitions, we construct a differentiable episode rollout and analytically compute policy gradients from episode-level losses, improving both training efficiency and policy performance. After approximately 30 minutes of training on a single GPU, APG-Planner demonstrates comparable or superior performance to existing baseline methods in simulation and achieves speeds of up to 11\,m/s in real-world flight experiments. We further validate the proposed framework on a skid-steer ground robot, demonstrating its applicability to differentially flat robotic systems beyond quadrotors.

Related papers

Publication record

Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.

First announced by arXiv: 2026-11-13 · Indexed by Pub-AI: 2026-11-13 UTC

preprint

APG-Planner: Learning Vision-Based High-Speed Trajectory Planning via Analytic Policy Gradient

Jinwoo Lee, Geunsik Bae, Changseung Kim, Jiwoo Kim, Woojae Shin, Hyondong Oh

First announced by arXiv: 2026-11-13 · Original record ↗

arxiv
2611.15296
Explicit source versions (1)
  • v1 · Date not supplied
  1. Preprint available

    Source date: 2026-11-13 · First detected: 2026-11-13 UTC · Historical baseline

    Source evidence (1)
    • Official evidence ↗ · RSS guid=oai:arXiv.org:2611.15296v1; arxiv:announce_type=new; item pubDate (official metadata)
Abstract · arxiv

Vision-based high-speed navigation requires a trajectory planner that can reactively generate smooth, collision-free trajectories from onboard visual observations. Deep reinforcement learning offers a promising approach to vision-based trajectory planning, as it enables the optimization of sequential decision-making policies in complex environments. However, model-free policy gradient methods rely on sample-based policy gradient estimation, which yields noisy gradient estimates and inefficient training. To this end, we propose a framework for learning trajectory planning policies using the analytic policy gradient (APG) method and use it to train APG-Planner, a vision-based high-speed trajectory planner. The proposed framework leverages differential flatness to analytically reconstruct vehicle states from planned trajectories, thereby enabling differentiable trajectory-level state transitions without requiring an explicit low-level dynamics model. By iteratively performing trajectory planning and differentiable state transitions, we construct a differentiable episode rollout and analytically compute policy gradients from episode-level losses, improving both training efficiency and policy performance. After approximately 30 minutes of training on a single GPU, APG-Planner demonstrates comparable or superior performance to existing baseline methods in simulation and achieves speeds of up to 11\,m/s in real-world flight experiments. We further validate the proposed framework on a skid-steer ground robot, demonstrating its applicability to differentially flat robotic systems beyond quadrotors.