HERON: Real-Time Obstacle-Aware Reachability Maps for Mobile Manipulators
Overview
HERON is a 3D U-Net trained to predict obstacle-aware reachability maps at 2 cm resolution from live depth data, enabling real-time operation on embedded hardware. It achieves an average prediction time of 115 ms with an 8.57 Hz update rate, outperforming the 90-minute computation time of cuRobo. The model demonstrates high accuracy with Pearson correlation of 0.99 and quadratic weighted kappa above 0.9.
From arXiv Robotics — research abstracts
Event evidence and corrections
0 attributed source owners. Ownership does not establish independent confirmation. Quantities are reported separately and are never added together.
Reported quantity · resolution: 2 other · Basis not reported
“obstacle-aware RM at 2 cm resolution from a live depth camera, on embedded hardware, fast enough for online use. We present HERON, a 3D U-Net trained on RMs computed with the GPU-accelerated inverse kinematics of cuRobo, first on synthetic scenes and then on real depth data, and deployed on a Jetson AGX Orin with TensorRT and INT8 quantization. HERON predicts the RM of a 2.56 m cube around the arm in 115 ms on avera”
Exact source · revision 1Source owner not reported
Reported quantity · prediction time: 115 other · Basis not reported
“obstacle-aware RM at 2 cm resolution from a live depth camera, on embedded hardware, fast enough for online use. We present HERON, a 3D U-Net trained on RMs computed with the GPU-accelerated inverse kinematics of cuRobo, first on synthetic scenes and then on real depth data, and deployed on a Jetson AGX Orin with TensorRT and INT8 quantization. HERON predicts the RM of a 2.56 m cube around the arm in 115 ms on avera”
Exact source · revision 1Source owner not reported
Reported quantity · voxel accuracy: 95 other · Basis not reported
“ge, for a measured system rate of 8.57 Hz, close to the 10 Hz update rate of the obstacle map, whereas computing the same map with cuRobo takes about 90 min. Against the cuRobo ground truth, 95% of the voxels lie within 1 class out of 51 with the FP32 model and within 3 classes with the INT8 model, with a Pearson correlation of 0.99 and a quadratic weighted kappa above 0.9. Accuracy degrades mainly at the workspace b”
Exact source · revision 1Source owner not reported
Reported quantity · Pearson correlation: 0.99 other · Basis not reported
“ge, for a measured system rate of 8.57 Hz, close to the 10 Hz update rate of the obstacle map, whereas computing the same map with cuRobo takes about 90 min. Against the cuRobo ground truth, 95% of the voxels lie within 1 class out of 51 with the FP32 model and within 3 classes with the INT8 model, with a Pearson correlation of 0.99 and a quadratic weighted kappa above 0.9. Accuracy degrades mainly at the workspace b”
Exact source · revision 1Source owner not reported
Reported quantity · quadratic weighted kappa: 0.9 other · Basis not reported
“ge, for a measured system rate of 8.57 Hz, close to the 10 Hz update rate of the obstacle map, whereas computing the same map with cuRobo takes about 90 min. Against the cuRobo ground truth, 95% of the voxels lie within 1 class out of 51 with the FP32 model and within 3 classes with the INT8 model, with a Pearson correlation of 0.99 and a quadratic weighted kappa above 0.9. Accuracy degrades mainly at the workspace b”
Exact source · revision 1Source owner not reported
Reported quantity · map computation time: 90 other · Basis not reported
“ge, for a measured system rate of 8.57 Hz, close to the 10 Hz update rate of the obstacle map, whereas computing the same map with cuRobo takes about 90 min. Against the cuRobo ground truth, 95% of the voxels lie within 1 class out of 51 with the FP32 model and within 3 classes with the INT8 model, with a Pearson correlation of 0.99 and a quadratic weighted kappa above 0.9. Accuracy degrades mainly at the workspace b”
Exact source · revision 1Source owner not reported
Evidence dependency changes (1)
- 2026-10-31T04:07:43.069Z · evidence updated · source revision 1. Evidence extraction was updated; current source attributions are shown above.
Report timeline
Follow attributed reports and material updates.
- arXiv Robotics — research abstractsSignalHERON: Real-Time Obstacle-Aware Reachability Maps for Mobile Manipulators
HERON is a 3D U-Net trained to predict obstacle-aware reachability maps at 2 cm resolution from live depth data, enabling real-time operation on embedded hardware. It achieves an average prediction time of 115 ms with an 8.57 Hz update rate, outperforming the 90-minute computation time of cuRobo. The model demonstrates high accuracy with Pearson correlation of 0.99 and quadratic weighted kappa above 0.9.
Event coverage history
There is not enough continuous observation data to show a trend.
Timezone · UTC
Article dates follow your selected timezone. Briefing editions use Hong Kong time (UTC+8).