English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Many autonomous systems forecast aspects of the future in order to aiddecision-making. For example, self-driving vehicles and robotic manipulationsystems often forecast future object poses by first detecting and trackingobjects. However, this detect-then-forecast pipeline is expensive to scale, aspose forecasting algorithms typically require labeled sequences of objectposes, which are costly to obtain in 3D space. Can we scale performance withoutrequiring additional labels? We hypothesize yes, and propose inverting thedetect-then-forecast pipeline. Instead of detecting, tracking and thenforecasting the objects, we propose to first forecast 3D sensor data (e.g.,point clouds with $100$k points) and then detect/track objects on the predictedpoint cloud sequences to obtain future poses, i.e., a forecast-then-detectpipeline. This inversion makes it less expensive to scale pose forecasting, asthe sensor data forecasting task requires no labels. Part of this work's focusis on the challenging first step -- Sequential Pointcloud Forecasting (SPF),for which we also propose an effective approach, SPFNet. To compare ourforecast-then-detect pipeline relative to the detect-then-forecast pipeline, wepropose an evaluation procedure and two metrics. Through experiments on arobotic manipulation dataset and two driving datasets, we show that SPFNet iseffective for the SPF task, our forecast-then-detect pipeline outperforms thedetect-then-forecast approaches to which we compared, and that pose forecastingperformance improves with the addition of unlabeled data.

Inverting the Pose Forecasting Pipeline with SPF2: Sequential Pointcloud Forecasting for Sequential Pose Forecasting