Papers › SoFar: Language-Grounded Orientation Bridges Spatial Reasoning and Object Manipulation

SoFar: Language-Grounded Orientation Bridges Spatial Reasoning and Object Manipulation

18 Feb 2025arXiv:2502.13143archive 2025-07-28

Zekun Qi, Wenyao Zhang, Yufei Ding, Runpei Dong, Xinqiang Yu, Jingwen Li, Lingyun Xu, Baoyu Li, Xialin He, Guofan Fan, Jiazhao Zhang, JiaWei He, Jiayuan Gu, Xin Jin, Kaisheng Ma, Zhizheng Zhang, He Wang, Li Yi

Spatial intelligence is a critical component of embodied AI, promoting robots to understand and interact with their environments. While recent advances have enhanced the ability of VLMs to perceive object locations and positional relationships, they still lack the capability to precisely understand object orientations-a key requirement for tasks involving fine-grained manipulations. Addressing this limitation not only requires geometric reasoning but also an expressive and intuitive way to represent orientation. In this context, we propose that natural language offers a more flexible representation space than canonical frames, making it particularly suitable for instruction-following robotic systems. In this paper, we introduce the concept of semantic orientation, which defines object orientations using natural language in a reference-frame-free manner (e.g., the ''plug-in'' direction of a USB or the ''handle'' direction of a knife). To support this, we construct OrienText300K, a large-scale dataset of 3D models annotated with semantic orientations that link geometric understanding to functional semantics. By integrating semantic orientation into a VLM system, we enable robots to generate manipulation actions with both positional and orientational constraints. Extensive experiments in simulation and real world demonstrate that our approach significantly enhances robotic manipulation capabilities, e.g., 48.7% accuracy on Open6DOR and 74.9% accuracy on SIMPLER.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

qizekun/SoFar mentioned on GitHubpytorch report
zhangwenyao1/open6dor_v2_execution mentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Object RearrangementRobot ManipulationRobot NavigationSpatial ReasoningVisual Question Answering

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Object Rearrangement Open6DOR V2 SoFar 6-DoF 48.7 #1 of 5 Archive leaderboard report
Object Rearrangement Open6DOR V2 SoFar pos-level0 96.0 #1 of 5 Archive leaderboard report
Object Rearrangement Open6DOR V2 SoFar pos-level1 81.5 #1 of 5 Archive leaderboard report
Object Rearrangement Open6DOR V2 SoFar rot-level0 68.6 #1 of 5 Archive leaderboard report
Object Rearrangement Open6DOR V2 SoFar rot-level1 42.2 #1 of 5 Archive leaderboard report
Object Rearrangement Open6DOR V2 SoFar rot-level2 70.1 #1 of 5 Archive leaderboard report
Robot Manipulation SimplerEnv-Google Robot SoFar Variant Aggregation 0.676 #1 of 9 Archive leaderboard report
Robot Manipulation SimplerEnv-Google Robot SoFar Variant Aggregation-Move Near 0.740 #1 of 9 Archive leaderboard report
Robot Manipulation SimplerEnv-Google Robot SoFar Variant Aggregation-Open/Close Drawer 0.297 #1 of 9 Archive leaderboard report
Robot Manipulation SimplerEnv-Google Robot SoFar Variant Aggregation-Pick Coke Can 0.907 #1 of 9 Archive leaderboard report
Robot Manipulation SimplerEnv-Google Robot SoFar Visual Matching 0.749 #1 of 9 Archive leaderboard report
Robot Manipulation SimplerEnv-Google Robot SoFar Visual Matching-Move Near 0.917 #1 of 9 Archive leaderboard report
Robot Manipulation SimplerEnv-Google Robot SoFar Visual Matching-Open/Close Drawer 0.403 #1 of 9 Archive leaderboard report
Robot Manipulation SimplerEnv-Google Robot SoFar Visual Matching-Pick Coke Can 0.923 #1 of 9 Archive leaderboard report
Robot Manipulation SimplerEnv-Widow X SoFar Average 0.583 #1 of 7 Archive leaderboard report
Robot Manipulation SimplerEnv-Widow X SoFar Put Carrot on Plate 0.667 #1 of 7 Archive leaderboard report
Robot Manipulation SimplerEnv-Widow X SoFar Put Eggplant in Yellow Basket 0.375 #1 of 7 Archive leaderboard report
Robot Manipulation SimplerEnv-Widow X SoFar Put Spoon on Towel 0.583 #1 of 7 Archive leaderboard report
Robot Manipulation SimplerEnv-Widow X SoFar Stack Green Block on Yellow Block 0.708 #1 of 7 Archive leaderboard report
Spatial Reasoning 6-DoF SpatialBench SoFar Orientation-abs 31.3 #1 of 7 Archive leaderboard report
Spatial Reasoning 6-DoF SpatialBench SoFar Orientation-rel 54.6 #1 of 7 Archive leaderboard report
Spatial Reasoning 6-DoF SpatialBench SoFar Position-abs 33.8 #1 of 7 Archive leaderboard report
Spatial Reasoning 6-DoF SpatialBench SoFar Position-rel 59.6 #1 of 7 Archive leaderboard report
Spatial Reasoning 6-DoF SpatialBench SoFar Total 43.9 #1 of 7 Archive leaderboard report
Spatial Reasoning EmbSpatial-Bench SoFar Generation 70.88 #1 of 5 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections