en.Wedoany.com Reported - On June 3, Daimon Robotics, in collaboration with Galaxy General, officially released RobOmni during ICRA, a full-modal evaluation benchmark for physical interaction capabilities that includes tactile sensing. RobOmni focuses on contact-rich operation scenarios, providing a standardized set of test tasks for fine operations such as peg-in-hole insertion, workpiece assembly, and screw fastening.
The release of RobOmni addresses a critical gap in embodied intelligence, moving from "seeing and understanding" to "precise touch and stable operation." Current robot models have made rapid progress in visual recognition, language understanding, path planning, and basic grasping. However, when entering real-world scenarios such as industrial assembly, laboratory operations, retail restocking, medical assistance, and household services, many tasks require more than just recognizing objects and moving them; they demand continuous adjustment of actions based on contact, force, friction, clearance, posture, and material feedback. Tasks like peg-in-hole insertion, workpiece assembly, and screw fastening may seem minor, but they require robots to perceive contact states with minimal error and correct operation paths based on tactile feedback. By incorporating tactile information into a full-modal evaluation framework, RobOmni enables different robot bodies, grippers, dexterous hands, and operation models to be compared under the same task protocol, reducing evaluation bias caused by relying solely on visual or single success rate metrics.
RobOmni has currently completed a 1:1 digital twin of the DM-TacClaw tactile gripper and will later expand to a five-fingered dexterous hand configuration. This progress means the evaluation system is not limited to a single gripper but is prepared to extend to more complex robot hand forms closer to real operational needs.
For the robotics industry, standardized evaluation benchmarks have infrastructure significance. Embodied intelligence companies are simultaneously advancing large models, motion control, end effectors, data collection, simulation training, and real-world validation. However, without reproducible, comparable, and scalable evaluation entry points, it is difficult to form a unified judgment of algorithmic and hardware capabilities. RobOmni connects the digital twin of tactile grippers, contact-rich task sets, and subsequent dexterous hand extensions, providing a clearer testing environment for tactile sensors, robot operation models, reinforcement learning strategies, and simulation-to-reality transfer. Daimon Robotics focuses on high-resolution multimodal tactile perception, tactile dexterous hands, and vision-tactile-language operation models, while Galaxy General targets general embodied intelligence bodies and scenario deployment. The joint release of RobOmni by the two companies reflects that tactile operation capabilities are moving from single-point hardware innovation into systematic evaluation and engineering validation stages.
Future variables center on RobOmni's open scope, task set expansion speed, five-fingered dexterous hand adaptation progress, real-world validation entry points, and industry developer participation. As robot applications shift from demonstrative grasping to high-precision assembly, flexible manufacturing, and real service scenarios, tactile evaluation benchmarks will become an important tool for determining whether robots can stably complete complex physical tasks. If RobOmni can form a continuously iterating open evaluation system, it will help accelerate collaborative validation among tactile operation models, end effectors, and embodied intelligence applications.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
