A team of researchers from the University of Illinois Urbana-Champaign, Carnegie Mellon University, the Georgia Institute of Technology, the University of California at Berkeley, Meta AI Research and Mistral AI developed a navigation system universal called GO To Any Thing (GOAT). This system is designed for extended standalone operation in domestic and warehouse environments. GOAT is a multimodal system that can interpret goals from category labels, target images, and linguistic descriptions. This is a permanent system that benefits from past experiences. GOAT is platform independent and adaptable to various robot embodiments.
GOAT, a general-purpose navigation system for mobile robots, is adept at autonomous navigation in various environments using category labels, target images, and linguistic descriptions. GOAT uses depth estimations and semantic segmentation to create a 3D semantic voxel map for accurate object instance detection and memory storage. The semantic map facilitates spatial representation, tracking of object instances, obstacles and explored areas.
GOAT is a mobile robotic system inspired by knowledge of animal and human navigation. GOAT, a universal navigation system, operates autonomously in various environments, performing tasks based on human input. Multimodal, persistent, and platform-independent, GOAT uses category labels, target images, and language descriptions for goal specification. The study evaluates the performance of GOAT in reaching unseen multimodal object instances and highlights its superiority, leveraging SuperGLUE-based image keypoint matching compared to CLIP feature matching in methods previous ones.
GOAT, a universal navigation system, uses a modular design and instance-aware semantic memory for multimodal navigation based on images and linguistic descriptions. The plan, platform-independent and capable of lifelong learning, demonstrates its capabilities through large-scale real-world experiments in homes. Using metrics such as success weighted by path length, GOAT’s performance is evaluated without precomputed maps. The agent uses global and local policies, using Fast Marching for path planning and point navigation controllers to reach waypoints along the path.
In experimental trials in nine homes, GOAT, a universal navigation system, achieved a success rate of 83%, outperforming previous methods by 32%. His success rate increased from 60% on the first objective to 90% after the exploration, demonstrating his adaptability. GOAT seamlessly handled downstream tasks like pick and place and social navigation. Qualitative experiments demonstrated the deployment of GOAT on the Boston Dynamics Spot and Hello Robot Stretch robots. Large-scale quantitative experiments on Spot in real homes highlighted GOAT’s superior performance against three benchmarks, outperforming matched instances and efficient navigation.
An exceptional, platform-independent multimodal design allows goals to be specified through a variety of means, including category labels, target images, and language descriptions. Modular architecture and instance-aware semantic memory distinguish instances of the same category for efficient navigation. Evaluated in large-scale experiments without precomputed maps, GOAT demonstrates versatility, extending its capabilities to tasks such as choice and placement and social navigation.
The future trajectory of GOAT involves a comprehensive exploration of its performance in various environments and scenarios to assess its generalizability and robustness. The surveys will aim to improve the match threshold to respond to the difficulties encountered during prospecting. Downsampling instances based on objective category will be further investigated to improve performance. Continued development of GOAT includes refinement of global and local policies and potential integration of additional techniques for more efficient navigation. An in-depth real-world evaluation will encompass different robots and tasks to validate GOAT’s versatility. Further exploration may extend the applicability of GOAT beyond navigation to areas such as object recognition, manipulation, and interaction.
Check Paper And Project. All credit for this research goes to the researchers of this project. Also don’t forget to register our SubReddit 33k+ ML, 41,000+ Facebook communities, Discord Channel, And E-mailwhere we share the latest AI research news, interesting AI projects and much more.
If you like our work, you will love our newsletter.
Hello, My name is Adnan Hassan. I’m a consulting intern at Marktechpost and soon to be a management intern at American Express. I am currently pursuing a dual degree at Indian Institute of Technology, Kharagpur. I am passionate about technology and want to create new products that make a difference.
