基於物理人工智慧的模型：追蹤器

具體益處：

• 確定與您的機器人產品和平台策略最相關的實體人工智慧（AI）模型、開發人員和技術方法。
• 比較模型許可、部署架構、硬體需求、相容的機器人類型和關鍵用例，以協助進行技術評估和夥伴關係決策。
• 在評估機器人基礎模型時，我們會評估模擬資料和現實世界效能資料的可用性和限制。
• 我們追蹤快速發展的實體人工智慧生態系統中的關鍵模型發布、商業部署、資金籌措趨勢和策略部署。

重要問題的答案：

• 有哪些物理人工智慧平台模型可用於人形機器人、機械手臂、移動機器人、自動駕駛汽車和其他具身系統？
• 主要模型在模型類型、許可方式、參數數量、部署架構、硬體平台和預期用例方面有何不同？
• 哪些模型已公佈了在模擬或真實環境中的成功率？比較這些結果時有哪些限制？

調查要點：

• 對機器人製造商 (OEM)、人工智慧公司、技術供應商、新創公司和學術聯盟開發的實體人工智慧平台模型進行詳細追蹤調查。
• 這涵蓋了模型版本、許可、模型類型、發布日期、模擬和實際成功率、參數數量、支援的機器人類型、邊緣和雲端部署、硬體平台、用例、部署範例、資金籌措和季度趨勢。
• 分析人形機器人、單臂和雙臂機械手臂、移動機械手臂、自動駕駛車輛及其他表現形式的模型。

本報告的主要目標受眾

• 一位在機器人OEM廠商中擔任產品、策略和人工智慧領域領導者的人士，正在評估當前和未來機器人平台的基礎模型。
• 一家技術供應商，致力於為實體人工智慧工作負載開發運算、半導體、模擬、雲端運算、邊緣人工智慧、感測器和機器人軟體解決方案。
• 企業發展、創業投資和投資領域的專家正在評估開發實體人工智慧的公司的競爭格局和商業性成熟度。

目錄

圖表

• 圖 1：每個模型的參數數量
• 圖 2 各模型在實際環境的報告成功率
• 圖 3. 按模型報告的模擬成功率。

Actionable Benefits:

• Identify the Physical Artificial Intelligence (AI) models, developers, and technology approaches most relevant to robotics product and platform strategies.
• Compare model licensing, deployment architecture, hardware requirements, supported robot types, and primary use cases to inform technology evaluation and partnership decisions.
• Assess the availability and limitations of simulation and real-world performance data when evaluating foundation models for robotics.
• Track significant model releases, commercial deployments, funding activity, and strategic developments across the rapidly evolving Physical AI ecosystem.

Critical Questions Answered:

• Which Physical AI foundation models are available for humanoids, robotic manipulators, mobile robots, autonomous vehicles, and other embodied systems?
• How do leading models differ by model type, licensing approach, parameter count, deployment architecture, hardware platform, and intended use case?
• Which models have published simulation or real-world success rates, and what are the limitations of comparing these results?

Research Highlights:

• A detailed tracker of Physical AI foundation models developed by robotics Original Equipment Manufacturers (OEMs), AI companies, technology vendors, startups, and academic consortia.
• Coverage of model version, licensing, model type, release date, simulation and real-world success rates, parameter count, supported robot types, edge and cloud deployment, hardware platform, use case, deployments, funding, and quarterly developments.
• Analysis of models supporting humanoids, single-arm and dual-arm manipulators, mobile manipulators, autonomous vehicles, and other embodied systems.

Who Should Read This?

• Robotics OEM product, strategy, and AI leaders evaluating foundation models for current and future robot platforms.
• Technology providers developing compute, semiconductor, simulation, cloud, edge AI, sensor, and robotics software solutions for Physical AI workloads.
• Corporate development, venture capital, and investment professionals assessing the competitive landscape and commercial maturity of Physical AI developers.

TABLE OF CONTENTS

Tables

• Table 1 Physical AI Foundation Model Tracker

Charts

• Chart 1 Parameter Count by Model
• Chart 2 Reported Real-World Success Rate by Model
• Chart 3 Reported Simulation Success Rate by Model