AI robots solve the pain point of tennis ball collection, the Haimoxin team seeks media coverage

During tennis training, players often spend nearly half their time bending down to pick up balls. Traditional ball machines can only serve balls, lacking the ability to autonomously locate and retrieve tennis balls, which makes it difficult to form a coherent training closed loop. Shenzhen Haimoxin Technology Co., Ltd. (hereinafter referred to as "Haimoxin Technology") has independently developed the NovaAceRobot intelligent tennis robot targeting this scenario, integrating AI visual recognition, autonomous navigation, and intelligent serving into a single mobile platform. At present, the third-generation functional prototype has completed validation on real tennis courts, and the Haimoxin team is preparing for the global Kickstarter crowdfunding launch.

 I. The Efficiency Bottleneck in Tennis Training Spawns the Integrated "Ball Picking + Serving" Demand

Tennis maintains a stable enthusiast base and training market worldwide. For amateur players, youth training camps, and private coaches, the cycle of "serving - hitting - picking up balls" during solo practice is the main bottleneck restricting training intensity and concentration. Existing commercial ball machine products mainly serve balls from fixed positions, unable to autonomously patrol the court, identify scattered tennis balls, and complete collection, forcing users to frequently interrupt training to pick up balls manually.

Meanwhile, sectors such as intelligent sports technology and consumer-grade service robots continue to heat up. The market has a clear demand for embodied intelligent products that can function in real venues rather than remaining at the conceptual stage. How to deeply integrate computer vision, mobile robot navigation, and sports training scenarios, and implement them in an affordable consumer-grade form, has become the core proposition of this niche field.

 II. Full-Stack Self-Developed NovaAceRobot, Closing the "Recognition - Navigation - Ball Picking - Serving" Loop

Haimoxin Technology was founded in 2024, headquartered in Shenzhen, with a technical roadmap of "AI visual perception + intelligent decision-making + autonomous mobile robot". Its flagship product, NovaAceRobot (internal codename Nova), is positioned as an AI tennis robot integrating automatic ball collection, intelligent serving, SLAM navigation, and video recording.

In terms of technical architecture, the Haimoxin team has built a complete ROS2 software stack, integrating Nav2 autonomous navigation, LiDAR SLAM mapping, and IMU fusion positioning, enabling the robot to perform autonomous patrol, path planning, dynamic obstacle avoidance, and precise target approaching on the tennis court. On the visual front, the team developed a tennis target recognition and multi-target tracking system based on the YOLO deep learning framework, deployed on the NVIDIA Jetson edge computing platform to achieve real-time inference under low power consumption. The entire machine adopts a LiDAR and vision fusion perception solution, supports high-precision indoor and outdoor positioning, and has the capacity to pick up up to 3 balls simultaneously.

Regarding product features, the standard ball bin of NovaAceRobot has a capacity of 60 to 80 balls, which can be upgraded to over 150 balls; its serving speed ranges from 30 to over 100 km/h, supporting topspin, backspin, and no-spin modes, with a serving interval of approximately 4 to 10 seconds per ball. Users can perform multi-modal operations via iOS/Android App or remote controller to adjust parameters, complete mapping and navigation, and play back or share training videos. The whole machine weighs about 16.75 kg, with dimensions of 0.36×0.5×0.75 meters, a battery life of 3 to 5 hours, and supports about 2 hours of fast charging.

Compared with single-function ball machines on the market, the core difference of NovaAceRobot is that it integrates "ball finding, ball picking, serving, and recording" into a single autonomous mobile unit. Users do not need to bend down repeatedly to pick up balls during training, allowing them to spend more time resting to replenish physical energy, resulting in a more coherent training rhythm and a more relaxed sports experience.

 III. Targeting Individual Users and Training Scenarios, Entering the Global Market via Crowdfunding

The target users of Haimoxin Technology mainly include tennis enthusiasts, private coaches, youth training camps, and high-end home sports scenarios. The product adopts a hardware sales model, with plans to debut in the overseas consumer market through Kickstarter, targeting a crowdfunding sales volume of 1000 units to complete the first batch of mass production and supply chain validation accordingly. The team has finalized the BOM cost structure and connected with the supply chain to prepare for the first batch of mass production after the crowdfunding goal is met. The company is simultaneously advancing patent layout, having applied for national patents, and planning a number of invention, utility model, and design patents to provide intellectual property support for subsequent large-scale production and global brand expansion.

In terms of market space, the global tennis participant base is considerable, and the consumption upgrading trend of smart sports equipment is obvious; domestic sports training and national fitness policies continue to drive demand for venues and equipment. Haimoxin Technology chooses tennis robots as its first landing scenario, aiming to verify the "AI vision + autonomous navigation" technology stack in a vertical scenario, and accumulate engineering experience for subsequent expansion into directions such as multi-robot collaboration, cloud scheduling, and embodied intelligence.

IV. From Independent Engineer Breakthrough to Team Collaboration, the Prototype Has Operated Successfully on Real Courts

The R&D journey of Haimoxin Technology began with the personal practice of founder Xin Huanrong. With 15 years of background in mechanical structure design, Xin Huanrong realized the inefficiency of the ball-picking process while learning tennis on his own in 2023, and immediately decided to independently develop a robot that can patrol the court, identify, approach, and pick up balls. Over the following two years, Xin Huanrong independently conquered technical modules including ROS2, SLAM, and YOLO, 3D printed and iterated over 445 parts in total, and spent more than 730 nights completing full-stack development from mechanical structure, embedded control to algorithm integration.

In 2025, the third-generation prototype successfully ran core functions on a real tennis court, and the project transitioned from "individual breakthrough" to team collaboration: after the visual lead joined, the YOLO-based tennis recognition and multi-target tracking capabilities were enhanced; after the product lead joined, brand building, crowdfunding storytelling, and marketization preparation were advanced. At present, the team has completed functional prototype verification, with actual tests covering core capabilities such as visual ball picking, automatic serving, court navigation, and App/controller control, and retains on-site court test videos from June 2026 as product validation evidence. The company's official website, Chuangyezhi brand page, and crowdfunding review materials are all ready. The next phase of work will focus on Kickstarter launch preparation, including main video production, platform review, pre-launch promotion, and first batch mass production stocking. The Haimoxin team stated that the current product has moved past the "proof of concept" stage, and the next priority is to complete market validation and the prelude to scaling through global crowdfunding channels.