Researchers have developed SPARC, a new method for motion forecasting that estimates uncertainty in a single pass. This approach combines Bayesian and conformal methods to produce structured, calibrated, and efficient uncertainty estimates. The method uses a deterministic neural network backbone and a conjugate Bayesian last layer to predict the future mean and epistemic scale of motion forecasts. SPARC outperforms existing methods on several datasets and can be used as a lightweight risk monitor. --- Why it matters: This matters for researchers in AI because it provides a more efficient and accurate way to estimate uncertainty in motion forecasting, which is crucial for reliable deployment in applications such as robotics and autonomous vehicles. Source: https://arxiv.org/abs/2608.20802