# Practical Issues of Action-conditioned Next Image Prediction

arXiv: https://arxiv.org/abs/1802.02975 · HTML: https://mlsanity.com/paper/1802.02975/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 8 Feb 2018.

## Built on by (most cited first, 2 of 2)

- [Perceptual Motor Learning with Active Inference Framework for Robust Lateral Control](https://mlsanity.com/paper/2503.01676/) (2025)
- [Towards Human-Like Driving: Active Inference in Autonomous Vehicle Control](https://mlsanity.com/paper/2407.07684/) (2024)

## Abstract (arXiv)

The problem of action-conditioned image prediction is to predict the expected next frame given the current camera frame the robot observes and an action selected by the robot. We provide the first comparison of two recent popular models, especially for image prediction on cars. Our major finding is that action tiling encoding is the most important factor leading to the remarkable performance of the CDNA model. We present a light-weight model by action tiling encoding which has a single-decoder feedforward architecture same as [actionvideopredictionhonglak]. On a real driving dataset, the CDNA model achieves 0.3986 × 10-3 MSE and 0.9846 Structure SIMilarity (SSIM) with a network size of about bfseries 12.6 million parameters. With a small network of fewer than bfseries 1 million parameters, our new model achieves a comparable performance to CDNA at 0.3613 × 10-3 MSE and 0.9633 SSIM. Our model requires less memory, is more computationally efficient and is advantageous to be used inside self-driving vehicles.