2025 Academic Achievements Reviews: Top Universities Advance Robotics Research with Elephant Robotics' Cobots, AGVs and Humanoid Robots
With the mission of “Enjoy Robots World,” Elephant Robotics is dedicated to empowering global innovation by providing robots that combine reliable performance, an open-source ecosystem, scalable production, and cost-effectiveness. Rather than conducting scientific research itself, we focus on delivering practical, accessible robotic tools that help researchers bring their ideas to life—enabling them to build, test, and share breakthroughs more efficiently and accelerating progress across the global robotics and AI research community. Today, we are excited to look back on 2025 academic achievements enabled by Elephant Robotics’ robots and proudly present a pioneering collection of research featuring our collaborative robots (Cobots), automated guided vehicles (AGVs), and humanoid robots. Developed in partnership with leading institutions—including Cornell University, the University of Tokyo, Seoul National University, the University of Michigan, New York University, the University of Waterloo, and Waseda University—this collection highlights the innovative applications of our 6-DOF robotic arms—the myCobot and myPalletizer series—alongside the mobile robot myAGV and the semi-humanoid robot myBuddy 280. Together, these works showcase how Elephant Robotics’ robotic products are advancing real-world innovation across diverse fields in 2025.
In 2025, approximately 100 academic papers employed Elephant Robotics’ robots as core research tools. Together, these works demonstrate how robotics is accelerating progress across industry, agriculture, logistics, education, healthcare, commerce, and smart home services. This compilation serves as a comprehensive knowledge resource, capturing influential theoretical insights and practical applications while providing a clear view of the evolving landscape of human–robot collaboration. Targeted at robotics enthusiasts, educators, professionals, and researchers, this curated selection of 12 standout studies delivers an accessible and authoritative resource for tracking the newest innovations shaping robotics and its broader industrial applications.
Table of contents for this collection of papers:
- 1. Topic: In situ foliar augmentation of multiple species for optical phenotyping and bioengineering using soft robotics
- 2. Topic: Front Hair Styling Robot System Using Path Planning for Root-Centric Strand Adjustment
- 3. Topic: Dynamic inaudible frequency shifting communication for multi-robot collaboration in manufacturing
- 4. Topic: Trust Through Transparency: Explainable Social Navigation for Autonomous Mobile Robots via Vision-Language Models
- 5. Topic: Soft-Rigid Hybrid Revolute and Prismatic Joints Using Multilayered Bellow-Type Soft Pneumatic Actuators: Design, Characterization, and Its Application as Soft-Rigid Hybrid Gripper
- 6. Topic:Collaborative Heterogeneous Mini-Robotic 3D Printer for Manufacturing Complex Food Structures with Multiple Inks and Curved Deposition Surfaces
- 7. Topic: BestMan: Human–Robot Interaction Using Dynamic Hand Gesture for Teleoperation of Quadruped Robots with a Robotic Arm
- 8. Topic: Toward autonomous blackberry harvesting with a soft gripper and vision-controlled robotic arm
- 9. Topic: High flexibility of heterogeneous tri-robot collaborative handling
- 10. Topic: Find the Fruit: Designing a Zero-Shot Sim2Real Deep RL Planner for Occlusion Aware Plant Manipulation
- 11. Topic: Enhancing Healthcare Assistance with a Self-Learning Robotics System: A Deep Imitation Learning-Based Solution
- 12. Topic: AIRSPEED: An Open-source Universal Data Production Platform for Embodied Artificial Intelligence
1. Topic: In situ foliar augmentation of multiple species for optical phenotyping and bioengineering using soft robotics
Authors: Mehmet Mert Ilman, Annika Huber, Anand K. Mishra, Sabyasachi Sen, Fumin Wang, Tiffany Lin, Georg Jander, Abraham D. Stroock and Robert F. Shepherd
Universities: Cornell University, Manisa Celal Bayar University
Abstract: This study tackles major challenges in precision agriculture, where existing foliar delivery methods often damage leaves, work poorly on different plant species, and lack consistency. To solve this, the researchers created a soft robotic leaf gripper that can gently and effectively inject nanoparticles and genetic materials directly into plant leaves. The gripper was mounted on our 6 DOF robot arm myCobot 280 M5, which provided automatic positioning and enabled repeatable, stamp-based injections. Using the arm’s gravity-compensation teaching mode, the team easily demonstrated motions and automated the entire process. With this system, foliar delivery achieved a success rate of over 91% with minimal leaf damage, enabling reliable in vivo phenotyping and gene expression studies. This work shows how compact, affordable robotic arms can improve plant bioengineering, support high-throughput phenotyping, and push agricultural robotics toward more precise and automated solutions.
2. Topic: Front Hair Styling Robot System Using Path Planning for Root-Centric Strand Adjustment
Authors: Soonhyo Kim, Naoaki Kanazawa, Shun Hasegawa, Kento Kawaharazuka and Kei Okada
University: University of Tokyo
Abstract: This article introduces a robotic hair-styling system that uses image-based goal setting and root-centric strand adjustment to reproduce target hairstyles with high accuracy. The system is built using our 6-axis collaborative robotic arm myCobot 280 M5, equipped with a soft comb, which enables stable 3D trajectory execution and repeatable fine manipulation of hair strands. By integrating myCobot’s compact form factor and reliable motion control with a novel path-planning strategy, the system achieves precise alignment of hair strands with desired orientations, resulting in superior styling accuracy and consistency compared to random strand-selection baselines. The research demonstrates the ability of compact collaborative robots to improve automated grooming processes and catalyze emerging applications in service robotics and the beauty-tech sector.
3.Topic: Dynamic inaudible frequency shifting communication for multi-robot collaboration in manufacturing
Authors: Semin Ahn, Dohyeon Kim and Sung-Hoon Ahn
University: Seoul National University
Abstract: This paper introduces a dynamic inaudible frequency-shifting communication method that enables decentralized robot-to-robot interaction using 18–22 kHz sound signals. The system was validated using multiple heterogeneous robots, including our autonomous mobile robot myAGV and 6-DOF collaborative robotic arm myCobot 280 Pi, which served as Listener robots receiving and executing commands through the acoustic channel. The approach eliminates reliance on WiFi or Bluetooth networks and remains robust against noise and environmental interference. Experiments conducted with heterogeneous robots across one-to-one, one-to-two, and one-to-multiple configurations demonstrate communication accuracy above 97.5% at distances up to 4 meters, with only minimal performance loss in heavy-noise conditions. The results show that this method is a scalable, interference-resistant solution for real-time multi-robot collaboration, making it highly suitable for flexible manufacturing, autonomous factories, and other environments with limited network
4. Topic: Trust Through Transparency: Explainable Social Navigation for Autonomous Mobile Robots via Vision-Language Models
Authors: Oluwadamilola Sotomi, Devika Kodi and Aliasghar Arab
Universities: University of Michigan, New York University
Abstract: This paper presents a multimodal explainability framework that integrates Vision-Language Models (VLMs) and heat-map visualizations to enhance transparency during robot navigation. The system was tested on the myAGV during both manual and autonomous navigation, where an explainablility module was developed to detect social conflicts, generate visual reasoning cues, and provide natural-language explanations. Leveraging myAGV’s ROS-based mobility, onboard sensing, and real-time control, the framework effectively communicates the robot's intent and actions. In user studies with 30 participants, most preferred real-time explanations, reporting better understanding and greater confidence in the robot’s behavior. Confusion matrix analysis further confirmed the system’s accuracy and reliability. This research indicates that integrating explainability to AMRs substantially improves human-robot collaboration and improves usability in social environments.
5. Topic: Soft-Rigid Hybrid Revolute and Prismatic Joints Using Multilayered Bellow-Type Soft Pneumatic Actuators: Design, Characterization, and Its Application as Soft-Rigid Hybrid Gripper
Authors: Peter Seungjune Lee, Cameron Sjaarda, Run Ze Gao, Jacob Dupuis, Maya Rukavina-Nolsoe and Carolyn L. Ren
University: University of Waterloo
Abstract: This study presents the development of soft-rigid hybrid (SRH) joints using multilayered bellow-shaped soft pneumatic actuators (MBSPAs) to improve robotic system performance. The researchers developed both a revolute and a prismatic SRH joint and integrated them into a 3-point soft-rigid hybrid gripper mounted on the 4-axis robotic arm myPalletizer 260. This design addresses common limitations of conventional soft pneumatic actuators—such as low payload capacity and vulnerability to environmental damage—by enclosing them within rigid protective structures. Experimental results significantly improved force output and durability, highlighting the potential for agricultural automation, particularly in harvesting delicate fruits. This research not only advances the capabilities of soft robotics but also paves the way for greater efficiency and reliability in a variety of industrial applications.
6. Topic: Collaborative Heterogeneous Mini-Robotic 3D Printer for Manufacturing Complex Food Structures with Multiple Inks and Curved Deposition Surfaces
Authors: Karen Jazmin Mendoza-Bautista, Mariana S. Flores-Jimenez, Laisha Daniela Vázquez Tejeda Serrano, Grissel Trujillo de Santiago, Mario Moises Alvarez, Arturo Molina, Mariel Alfaro-Ponce and Isaac Chairez
Universities: Tecnológico de Monterrey, Nacional de México, CDMX
Abstract: This research presents a collaborative heterogeneous mini-robotic 3D printer developed to fabricate complex food structures using multiple inks and curved deposition surfaces. Incorporating the collaborative robot myCobot 280 M5 as part of the robotic manipulator, the printer effectively addresses challenges in multi-material food printing, such as extrusion consistency and the ability to create intricate geometries. The incorporation of this technology enables improved control over food ink rheology and facilitates the seamless integration of various materials, including synthetic meat and additives, for optimal texture and taste. The advancements showcased in this research significantly impact the food industry by promoting automation, reducing food waste, and offering tailored solutions for diverse dietary needs, thereby contributing to more sustainable food production practices.
7. Topic: Human–Robot Interaction Using Dynamic Hand Gesture for Teleoperation of Quadruped Robots with a Robotic Arm
Authors: Jianan Xie, Zhen Xu, Jiayu Zeng, Yuyang Gao and Kenji Hashimoto
University: Waseda University
Abstract: This article introduces an innovative human–robot interaction (HRI) system that uses dynamic hand gestures to teleoperate quadruped robots equipped with robotic arms. The system leverages the myCobot 280 Pi to enhance precision and intuitiveness in robot control. By integrating a Depth-MediaPipe framework for real-time 3D hand keypoint detection and a Semantic-Pose-to-Motion model, it accurately interprets gestures to perform complex robotic actions. Experiments with the Unitree Go1 and myCobot demonstrate smooth, stable, and highly accurate control performance. This technology offers strong potential for applications ranging from logistics to telemedicine, enabling more natural and efficient remote robot operation. The findings highlight the value of gesture-based control in improving user interaction and advancing teleoperated robotic systems for real-world use.
8. Topic: Toward autonomous blackberry harvesting with a soft gripper and vision-controlled robotic arm
Authors: Fabio Taddei Dalla Torre, Omar Faris, Philip H. Johnson and Marcello Calisti
Universities: University of Trento, University of Lincoln, Scuola Superiore Sant’Anna
Abstract: This research presents a novel framework for autonomous blackberry harvesting by using a low-cost 6-DOF cobot, specifically the myCobot 320 Pi, paired with a custom-designed soft inflatable gripper. The integration of YOLOv8 for vision-based control enables precise detection of ripe blackberries, enhancing the system's grasping effectiveness. Results indicate a peak vision accuracy of 98.4% and a noteworthy grasping success rate of 76.6%, despite some variability based on fruit orientation. This research demonstrates the potential of robotic harvesting system to improve operational efficiency in berry cultivation, paving the way for broader adoption of robotics in high-value crop industries. By addressing the challenges associated with traditional fruit harvesting methods, this study contributes significantly to the advancement of automated agricultural practices.
9. Topic: High flexibility of heterogeneous tri-robot collaborative handling
Authors: ZHANG Shuzhong, QI Chunyu, ZHANG Gong, SU Jiahong, QIU Weiqian and RUAN Yuzhen
Universities: Fujian University of Technology, South China University of Technology, Guangdong University of Technology and Education
Abstract: This paper addresses the challenges of compliance in collaborative handling tasks involving heterogeneous tri-robot systems. By employing the myCobot 280, the authors proposed a reinforcement learning control strategy based on Proximal Policy Optimization (PPO). A high-fidelity simulation environment was developed in the CoppeliaSim robotic simulator to compare the performance of traditional force control and the proposed RL approach. Results indicate that the RL method significantly enhances trajectory tracking accuracy and motion smoothness. The successful application of the myCobot 280 not only demonstrates improved synergy among the robots during collaborative tasks but also confirms the feasibility of leveraging advanced AI techniques in robotic solutions. This advancement has the potential to transform industrial automation by enabling more flexible and efficient operation in complex environments.
10. Topic: Find the Fruit: Designing a Zero-Shot Sim2Real Deep RL Planner for Occlusion Aware Plant Manipulation
Authors: Nitesh Subedi, Hsin-Jung Yang, Devesh K. Jha and Soumik Sarkar
University: Iowa State University
Abstract: This study focuses on the challenges of robotic manipulation in complex and cluttered agricultural environments, specifically targeting the task of fruit localization and occlusion resolution. Utilizing the dual-arm semi-humanoid robot myBuddy 280, the authors developed an end-to-end deep reinforcement learning (RL) framework capable of adeptly interacting with deformable plants. This approach enables the robot to uncover hidden fruits by learning to manipulate foliage without needing precise geometric modeling. The research demonstrates a significant advancement in automation for agricultural robotics, paving the way for scalable, perception-driven solutions that can operate effectively in dynamic and unpredictable settings, enhancing productivity and efficiency in the agricultural sector.
11. Topic: Enhancing Healthcare Assistance with a Self-Learning Robotics System: A Deep Imitation Learning-Based Solution
Authors: Yagna Jadeja, Mahmoud Shafik, Paul Wood and Aaisha Makkar
University: University of Derby
Abstract: This paper presents a Self-Learning Robotic System (SLRS) for healthcare assistance using Deep Imitation Learning (DIL). The SLRS solution can observe and replicate human demonstrations, thereby acquiring complex skills without the need for explicit taskspecific programming. Using the myCobot 280 Jetson Nano as the experimental platform, the system autonomously observes and mimics human actions to effectively assist healthcare professionals. By combining advanced perception with gesture recognition, it can perform complex tasks such as medication delivery and patient support, helping to address major operational inefficiencies. The implementation of this technology not only enhances the quality of patient care but also demonstrates the potential for robotics to transform the healthcare industry, promoting greater efficiency and adaptability in service delivery.
12. Topic: AIRSPEED: An Open-source Universal Data Production Platform for Embodied Artificial Intelligence
Authors: Xuan Xia, Bo Yu, Jialin Jiao, Xinmin Ding, Xing He, Haoran Tong, Yufei Lin, Tongyi Shen, Ning Ding and Shaoshan Liu
Universities: Shenzhen Institute of Artificial Intelligence and Robotics for Society, Dora-rs.ai
Abstract: To address the shortage and high cost of standardized training data for embodied AI, the authors introduce AIRSPEED—an open-source, hardware-agnostic platform that unifies real-world data collection, simulation-based generation, and automatic dataset assembly. Using the 6-axis cobot myCobot Pro 630 as a data collector, AIRSPEED compresses 62 MB/s of joint, gripper, and RGB-D streams into 2% keyframes and builds a ready-to-train pyramid dataset in under 30 ms per cycle. Benchmarks show it creates datasets 23.5× faster and generates synthetic data 7.7× faster than conventional manual workflows, while the full system costs under $5K. This study demonstrates that affordable, commercial robots can now support large-scale data pipelines, making high-quality embodied AI development more accessible to educators, SMEs, and the service-robot industry.
This curated collection showcases groundbreaking research that is accelerating progress in both scientific discovery and real-world applications. Spanning precision agriculture, soft robotics, explainable navigation, embodied AI, food manufacturing, healthcare assistance, and advanced teleoperation, these studies highlight the versatility and impact of our robotic ecosystem across a wide range of fields. By enabling reliable automation, intuitive interaction, and accessible experimentation, the works not only validate the performance of the myCobot, myPalletizer, myAGV, and myBuddy series but also demonstrate how affordable robotics can drive innovation worldwide. As the robotics landscape continues to evolve rapidly, Elephant Robotics remains committed to empowering researchers, educators, and industry leaders with the tools needed to shape the next generation of intelligent robotic solutions.
To view and download selected academic papers from 2025, please click the following link:
https://drive.google.com/drive/folders/1u6kNiEd_jq5-ZnQL3bzXs3xECkwAiYqK?usp=drive_link.