Table 4 Root-mean-squared error of ATE improvement for robust SLAM compared to ORB-SLAM2 on TUM datasets
From: A survey: which features are required for dynamic visual simultaneous localization and mapping?
| Â | Low-level SLAM | Use high-level in point-based SLAM | ||||
|---|---|---|---|---|---|---|
Point-based | Point-based | Point-line-based | ||||
Year | 2020 | 2020 | 2018 | 2018 | 2019 | 2019 |
References | Yang et al. [20] | Du et al. [22] | Besco et al. [26] | Yu et al. [28] | Cui and Ma [30] | Zhang et al. [41] |
s_static | 23.2% | – | – | 25.9% | 13.0% | 24.1% |
s_xyz | – | 18.2% | −66.7% | – | – | 3.1% |
s_rpy | – | – | – | – | – | −15.8% |
s_halfsphere | – | – | 15.0% | – | – | 58.6% |
w_static | 98.2% | 94.9% | 93.3% | 97.9% | 98.5% | 98.3% |
w_xyz | 97.5% | 95.6% | 96.9% | 96.7% | 97.5% | 97.7% |
w_rpy | 95.8% | 93.8% | 94.7% | 48.7% | 97.2% | 76.4% |
w_halfsphere | 95.4% | 92.7% | 92.9% | 93.76% | 95.0% | 96.7% |