中EN Submit a tool
Paper

Anthropic and Andon Labs Release Drone-Bench to Evaluate AI Models' Ability to Autonomously Operate Drones for Localization and Tracking Tasks

Published: Source: Anthropic: Research (Published Results · Webpage)

ShareXFacebookTelegramWhatsApp

Anthropic, in collaboration with Andon Labs, has launched Drone-Bench, a benchmark for testing AI models' ability to autonomously control quadrotor drones to locate and track designated individuals in indoor environments. The benchmark decomposes the task into five sub-tasks: 3D map reconstruction, localization, navigation, target detection, and following, enabling rapid evaluation through software simulation. Experiments show that the difficulty of this task chain is sufficient to distinguish models of different intelligence levels and reveals the trajectory of AI progress in physical world manipulation capabilities.

Read the original (opens in a new tab)

News stream data aggregated by AI HOT

Related newsLatest in this category
Paper Greg Brockman congratulates on solving a 40-year math problem
· X: Greg Brockman (@gdb)
Paper German Scientists Predict Chaotic Signals Using 400 Droplet Particles, Achieving About 1/10 the Accuracy of Memristor-Based Solutions
· IT Home (RSS)
Paper Alibaba Tongyi Qianwen Team Proposes Skill Self-Play Method
· X: Rohan Paul (@rohanpaul_ai)
Paper StateAct: A New Agent Approach for Long-Horizon Computer Tasks
· X: AK (@_akhaliq)
Paper Epoch AI Uses AI to Crack Galois Group Problem in 2-adic Number Fields
· X: Epoch AI (@EpochAIResearch)