When disaster strikes and lives are at risk, who can build the line of defense for humanity? When technology meets responsibility, how can machines reflect human values? ATEC 2025, themed "AI and Robotics Real-World Challenges", invites top universities, research institutions, tech companies, and geek teams from around the world to explore the use of technology for social good. The competition focuses on autonomous perception, cognitive reasoning, decision-making, autonomous navigation, dexterous manipulation, and virtual-physical migration to develop versatile embodied AI systems for complex scenarios. It features two tracks (AI Software Algorithms & Robot Hardware Design), including Online and Offline Competition, aimed at innovating participants to integrate advanced technologies in multiple fields, resolve technical difficulties, ultimately anchoring progress in human welfare.
Sign up now: https://www.atecup.com/competitions/atec2025
Registration: Feb 21, 2025 22:00 - Apr 10, 2025 10:00 (UTC+8）
Online Competition – Software Track: Mar 13, 2025 10:00 - May 8, 2025 10:00 (UTC+8）
Online Competition – Hardware Track: Feb 21, 2025 22:00 - Apr 25, 2025 10:00 (UTC+8）
Offline Competition: November 2025
This track focuses on a rescue mission, where Contestants must design and implement autonomous embodied AI agents. The robots must be able to autonomously perform perception, reasoning, decision-making, and route planning based on task clues. During execution, the robots must dynamically adjust to changes in the environment. The models submitted by the teams must demonstrate robustness and adaptability, and ensure stable operation in complex and changing environments.
The submitted AI agent models will be tested in a virtual environment to evaluate the contestants' comprehensive development skills and mastery of embodied intelligence technologies, such as perception, reasoning, and decision-making. This track consists of five difficulty levels (L1–L5). Challenges will be released in order based on the schedule. Contestants must complete each level to advance to the next. The final ranking and awards will be determined by their score in L5.
Registration: Feb 21, 2025 22:00 - Apr 25, 2025 10:00 (UTC+8) Online Competition: Mar 13, 2025 10:00 - May 8, 2025 10:00 (UTC+8) Submission of Code Review Materials: May 9, 2025 10:00 - May 15, 2025 10:00 (UTC+8) Code Review: May 16, 2025 - May 30, 2025 Result Announcement: Jun 2, 2025
For more information, refer to the Answering Guide.
This track consists of five levels (L1–L5) with each level scored separately. Scores do not carry over between levels. Only the final score in L5 will be used to determine awards.
The advancement rules are as follows:
- Each level has its own leaderboard, ranking contestants based on their scores.
- The advancement evaluation adopts a dynamic cutoff score mechanism, determined by the score distribution at the opening of each advancement stage (*A valid submission refers to a team submission that obtains an output score and it must be greater than 0).
- When a new level opens, all teams that meet the advancement threshold will automatically advance to this level.
The advancement times for each level are as follows (subject to minor adjustments as the competition progresses):
· L1 (Start Time): Mar 13, 2025 10:00 AM (The score line for advancement from L1 to L2 is 0.475.)
· L2: Mar 20, 2025 10:00 AM
· L3: Mar 27, 2025 10:00 AM
· L4: Apr 3, 2025 10:00 AM
· L5: Apr 17, 2025 10:00 AM
Notes:
- The time zone used in these rules is UTC+8. The system will automatically adjust the time in competition-related announcements based on your login location.
- When a level opens, the advancement threshold for this level is set based on the top 80% of teams with valid submissions.
- Once a level opens for advancement, any team that meets or exceeds the advancement threshold will advance automatically.
注：
- 本规则所述时区为UTC+8，请各参赛队伍阅读公告时注意时区换算，系统计时将依据IP登录地区时间自动换算；
- 晋级开放时，当前级别有效提交的前80%赛队所取得的分数将被定为该级别的晋级分数线；
- 开放晋级后，达到该级别晋级分数线的赛队也将自动晋级。
The Organizing Committee will provide computing credits to participating teams. For more details, please refer to the Rules for Issuance and Use of Computing Credits for ATEC 2025 Online Competition.
The Online Competition (Hardware Track) is designed to evaluate the locomotion and object manipulation capabilities of robot systems in unstructured terrains and environments. Teams must design and validate a system that integrates a legged robot with a robotic arm. While teams have the flexibility to choose their legged robot platform, they must independently design and construct the robotic arm, grasping system, and perception system.
The robot must demonstrate stable locomotion, adaptability to complex terrains, and precise object grasping and transportation in a simulated environment. Teams must submit a detailed design proposal, covering motion control algorithms, robotic arm structure, system integration interfaces, and safety and protection mechanisms. A simulation test is required to validate system stability, grasp accuracy, and impact resistance. Designs must consider weight distribution, dynamic loads, and vibrations of the robotic arm and mobile platform to ensure efficient performance in complex tasks.
Registration: Feb 21, 2025 22:00 - Apr 25, 2025 10:00 (UTC+8)
Online Competition: Feb 21, 2025 22:00 - Apr 25, 2025 10:00 (UTC+8)
Online Review: Apr 26, 2025 - May 30, 2025 10:00 (UTC+8)
Result Announcement: Jun 2, 2025 (UTC+8)
Teams must design and validate a system that integrates a legged robot with a robotic arm. Teams have the flexibility to select their legged robot, but they must independently design and construct the robotic arm, grasping system, and perception system.
The system must demonstrate advanced mechanical design and seamless system coordination when performing the assigned tasks. The following section describes the specific tasks:
- Mobile Platform Algorithm Design and Demonstration
a. Objective: Teams must develop an algorithm for either the officially provided quadruped or humanoid robot or a self-developed legged platform to ensure stable locomotion and adaptability to complex terrains such as slopes and obstacles within a simulation environment.
b. Algorithm Design Requirements:
· Provide a detailed report on the motion control algorithm and perception strategy.
· Conduct physics-based 3D simulations to assess robot stability under varying loads and terrain conditions.
· Ensure balanced weight distribution and stable multi-contact grasping after integrating with the robotic arm.
c. Demonstration Requirements:
· Independently create a multi-terrain simulation environment, with at least two challenging surfaces, such as stairs, slopes, rough terrain, or soft sand, to fully showcase the dynamic locomotion of the robot.
· Demonstrate the mobility of the robot in the simulation environment. - Robotic Arm and End Effector Design and Demonstration
a. Objective: Teams must design a robotic arm and end effector capable of grasping and transporting objects in coordination with the mobile platform.
b. Mechanical Design Requirements:
· Submit detailed design schematics for the robotic arm and end effector, specifying the joint design, linkage dimensions, actuation methods, and end effector type (such as grippers and suction cups).
· Analyze the motion range, structural rigidity, and payload capacity of the robotic arm and end effector. Conduct a simulation test to validate the stability and precision of the robotic arm in grasping tasks.
· Design a well-structured interface between the robotic arm and the mobile platform to ensure a compact and easy-to-service system.
c. Demonstration Requirements:
· Set up a scene for pick-and-place tasks by using basic geometric objects such as soda cans and irregular objects such as bananas to fully showcase the grasping and manipulation capabilities of the robot.
· Demonstrate high repeatability in grasping different objects within a simulation environment. - System Integration and Collaborative Operation
a. Objective: The robotic system must autonomously navigate to a target area, and the robotic arm must independently grasp, transport, and place objects as specified in the task.
b. System Integration Requirements：
· Provide a system assembly diagram and structural coordination scheme, detailing the interface design and connection methods between the mobile platform, robotic arm, and control module.
· Mitigate dynamic loads, vibrations, and other issues arising from the interaction between the mobile platform and robotic arm, and validate system stability through simulation.
· Provide a clear plan for implementing safety mechanisms, such as overload protection and collision mitigation structures.
c. Demonstration Requirements:
· Demonstrate a seamless robotic workflow by integrating the previous two tasks.
- Submission of Materials
a. Teams must submit all electronic materials, including a complete design document, simulation data, a video demonstration, and source code, via the designated platform before the specified deadline. b. The design document (PDF format) should cover the system architecture, mechanical design, control algorithm principles, and perception and decision-making strategies. The document should be logically structured and visually informative, clearly explaining design concepts, technical highlights, and innovations. The document should also include theoretical analysis, simulation results, and performance evaluation data.
c. Teams must provide detailed instructions for reproducing the simulation results, which include, but are not limited to, the following information:
· Environment Setup: List the required software, dependency libraries, and version details, along with step-by-step installation and configuration instructions.
· Code Execution Process: Explain the code structure, module functions, and execution process to ensure reviewers can easily reproduce the simulation results.
· Parameter Configuration and Debugging: Provide instructions on configuring and debugging key parameters to ensure the reproducibility and stability of the simulation.
· Result Validation and Data Analysis: Describe the method for validating system performance using simulation results and provide the data analysis methodology and conclusion. - Preliminary Screening
· Material Review: The Jury Committee will verify the integrity and format of all submissions to ensure that all required documents, such as design documents, simulation projects, and reproduction instructions, are included and meet the specified requirements.
· Screening Result: Teams will be evaluated based on the quality of their submissions, with the top teams advancing to the interview round. - Interview Evaluation
· Scoring Criteria: The Jury Committee will evaluate each submission based on the competition’s scoring criteria by reviewing design documents, video demonstrations, and reproduced simulation results. The evaluation focuses on two dimensions: “robot system design” and “innovation of the solution”.
· Scoring Guidelines: Clear guidelines and key points are provided for scoring in each dimension. Judges will submit scores via an online review system.
· Defense Presentation: The interview will be conducted online, where teams must provide a real-time simulation demonstration and respond to judges’ questions to showcase the technical highlights and innovations of their solutions. - Result Announcement
· Result Announcement: The competition results and team rankings will be officially published on the competition website and sent to teams via email to ensure transparency and fairness.
- Robot System Design (60 points)
Locomotion (20 points)
--
Evaluation Criteria:
Whether the simulation video demonstrates the stability and shock resistance of the legged robot, as ensured by its leg structure and gait design;
Whether the submitted simulation data provides sufficient evidence of the locomotion performance of the robot in the simulation environment for Task 1, including its adaptability to complex terrains and stability under load.
Evaluation Criteria:
Whether the robotic arm and end effector perform grasping and handling tasks smoothly, accurately, and reliably within the simulation environment;
Whether the design documentation clearly explain the principles behind key components, such as the gripping mechanism of the arm and transmission system, while providing solid theoretical support and validating the design through simulation.
Evaluation Criteria:
Whether the overall design incorporates a modular architecture, and whether the mechanical interfaces and inter-module connections are logically organized to facilitate assembly and maintenance;
Whether the energy and signal transmission scheme is robust and reliable, and whether the online simulation data and videos clearly demonstrate the overall coordination and task execution of the system;
Whether the robot successfully completes Task 1 and Task 2 in the simulation environment after full system integration, exhibiting efficiency, stability, and adaptability.
- Design Proposal (40 points)
Intelligence (5 points)
--
Evaluation Criteria:
Whether the design incorporates intelligent features, such as adaptive adjustments and environmental perception feedback, for basic navigation, posture adjustment, and grasping;
Whether the online simulation demonstrates effective intelligent feedback and adjustment, highlighting the ability of the system to make autonomous decisions in dynamic environments.
Evaluation Criteria:
Whether the mechanical structure, interface design, material selection, or safety features offer innovative solutions that surpass traditional design limitations;
Whether the design document and drawings present novel approaches to addressing industry challenges, supported by compelling evidence of innovation and validation results.
Evaluation Criteria:
Whether the design document is well-organized and visually informative to facilitate understanding and replication;
Whether the modular design, assembly process, and debugging tools are intuitive and user-friendly to ensure ease of use and maintenance;
Whether the document provides comprehensive installation instructions, user interface descriptions, and technical support to ensure the operability of the system and enhance the overall user experience.
Evaluation Criteria:
Whether the design includes adequate mechanical redundancy, cushioning mechanisms, and emergency shutdown protocols to ensure safety under abnormal conditions;
Whether the online simulation videos and data effectively demonstrate the ability of the system to activate protective mechanisms in response to sudden external forces or structural anomalies, thus safeguarding critical components and enabling rapid shutdown.
-
Design Document
Content --
Provide a detailed description of the overall mechanical structure of the robot system, including the modular integration plan, interface parameters for each module (mechanical, energy, and signal), and installation diagrams;
Include a clear explanation of the safety features, such as redundancy mechanisms and emergency shutdown systems that ensure reliability under abnormal conditions;
The design document should be well-organized, with clear illustrations and annotations for easy understanding and replication.
Format
--
Submit the document in PDF or Word format, with a clear layout, readable text, and labeled pictures and diagrams. All tables and diagrams must include explanations;
Name the file in the format of: TeamID_DesignDocument.pdf/docx. -
Simulation Data
Simulation Task
--
Demonstrate the performance of the robot in climbing stairs and grasping and transporting objects on a 3D physics simulation platform;
Submit three 2-5 minute simulation videos or GIFs, each showing one of the three sub-tasks (e.g., movement, grasping, or collaboration);
The videos should highlight the system's stability, precision, and collaboration, with brief descriptions.
The video format should be in the MP4 or GIF format, with at least 720p resolution;
Name the video files in the format of: TeamID_TaskX_DemoVideo.mp4/gif.
- File Submission
File Packaging and Size Limit
--
All files (design documentation, simulation videos, CAD models, robot models, and task execution code) must be bundled into a single compressed file, with a total size not exceeding 50MB;
Name the file in the format of: TeamID_SubmissionFiles.zip.
The submitted CAD models, simulation robot models, and task execution code must be complete and reproducible. The Jury Committee may conduct spot checks as needed;
Models and code should be accompanied by a brief explanatory document to ensure that the Jury can quickly understand and execute the models and code.
· Completeness of Materials: Teams must ensure all submitted materials (design documentation, simulation data, videos, and code) are complete, well-documented, and clearly presented for easy review by the judges.
· Online Presentation Criteria: The design should emphasize the mechanical structure, modular integration, and safety measures. The video should cover the entire task process, including responses to abnormal situations.
· Format Standard: All materials must follow a consistent format, with standardized file naming, to prevent any issues that could affect the review process.
· Time Limit: All materials must be uploaded before the deadline; late submissions will affect the score.
· Violation Rules:
A team shall be disqualified, if deemed to have committed the following actions intentionally.
a. The design and construction of the robot do not comply with the requirements of the competition rules.
b. Any act that poses danger to the game field, its surroundings, the robots, and/or people.
c. Any other act that goes against the spirit of fair competition.
d. Any act of non-compliance with the decisions of the Jury Committe.
The Offline Competition is related to the Online Competition but focuses on the comprehensive application of various skills and problem-solving abilities. Teams are required to tackle challenges based on the application scenarios provided during the competition.The Organizing Committee will release a list of required hardware equipment. Contestants may select and modify their equipment within the bounds of competition rules. Contestants are also encouraged to develop their own equipment.
After the Online Competition ends, the top 10 teams from each track will be invited to the Offline Competition, where they will form combined teams. Each combined team will consist of one Hardware Track team and one Software Track team.
Teams that register for both tracks and place in the top 10 for each track may choose one of the following options:
- Compete independently, occupying one qualification spot in each of the two tracks for entry to the Offline Competition.
- Join a combined team. If the rankings for the two tracks differ, the team will form a combined team under the track with the higher ranking by default. If the rankings for both tracks are the same, the team may select its preferred track to form a combined team within the time limit specified by the competition rules.
Additional Notes:
If a top-10 team declines to form a combined team, the next highest-ranked team will be invited to take its place. If a team withdraws from a combined team, the remaining team may select a new partner according to the above rules to participate in the Offline Competition.
As this is the first time the Competition is open to global participants, and to attract more outstanding teams, the Organizing Committee will invite teams, nominated by the Chair of the Jury Committee and selected through a vote based on competition rules, to participate in the Offline Competition. Up to five teams may be invited, with a maximum of eight members per team. These teams do not participate in the Online Competition. Instead, they advance directly to the Offline Competition and are not required to form combined teams.
Before the preparation period begins, the Organizing Committee will release a list of required hardware equipment. Contestants may select and customize their equipment in accordance with the competition rules. Contestants are also encouraged to develop their own innovative robotic systems for the Competition, with appropriate allowances provided. Further details on the equipment list and allowance usage will be available on the official ATEC website.
The offline competition will require each participating team to use their own robot system to complete the following 4 tasks in an outdoor environment:
- Object Sorting Task:
The robot needs to identify, pick up, transport, and classify specific types of items within a designated area. The system should be equipped with visual perception, precise manipulation, and classification decision-making capabilities. - Object Transport and Water Sprinkling Task:
The robot needs to locate target items within a designated area, to plan operation sequences for sprinklers, faucets and other equipment, and to achieve object transport and fixed-point operation tasks. The system shall integrate environmental perception, motion planning, precise control and other functional modules. - Field Endurance Task:
The robot needs to maintain stable movement across outdoor terrain, including arched bridges, stairs, and gentle slopes. The system should be equipped with multi-sensor data fusion and adaptive motion control capabilities. - Suspension Bridge Crossing Task:
The robot needs to navigate through unevenly-spaced suspension bridges, laying wooden planks autonomously to ensure successful crossing (if necessary). The system should be equipped with stable motion planning, visual perception, precise operation, and rapid transit capabilities.
The offline competition will provide participating teams with commercial-grade hardware supplies. Teams may choose independent research and development or configure their systems through the certified list provided by the Organizing Committee. Hardware specifications and configuration guidelines are scheduled for release in May 2025.