From 341b86aa771d31fed9db160dd225cd581c137c30 Mon Sep 17 00:00:00 2001 From: Vikrant Dewangan <44745884+Vikr-182@users.noreply.github.com> Date: Mon, 2 Oct 2023 22:02:04 +0530 Subject: [PATCH] Update README.md --- README.md | 63 +++++++++++++++++-------------------------------------- 1 file changed, 19 insertions(+), 44 deletions(-) diff --git a/README.md b/README.md index 232a87d..bb846b0 100644 --- a/README.md +++ b/README.md @@ -1,8 +1,12 @@ -

UAP-BEV (Uncertainty Aware Planning in Bird's Eye View) generated from Monocular Images

- - -

By Vikrant Dewangan1, Basant Sharma2, Sarthak Sharma1, Tushar Choudhary1, Aakash Aanegola1, -Arun Kumar Singh2, K. Madhava Krishna1

+

UAP-BEV: Uncertainty Aware Planning in Bird's Eye View representations

+ +[Vikrant Dewangan](https://vikr-182.github.io/) 1 +[Basant Sharma](https://www.etis.ee/CV/Basant_Sharma) 2 +[Sarthak Sharma](https://scholar.google.com/citations?user=4uKV9aIAAAAJ&hl=en) 1 +[Tushar Choudhary](https://tusharc31.github.io/)1 +[Aakash Aanegola](https://github.com/Aa-Aanegola)1, +[Arun Kumar Singh](https://scholar.google.co.in/citations?user=0zgDoIEAAAAJ&hl=en)2, +[K. Madhava Krishna](https://scholar.google.co.in/citations?user=QDuPGHwAAAAJ&hl=en) 1

(1- Robotics Research Center, IIIT Hyderabad, 2- University of Tartu, Estonia)

@@ -17,29 +21,11 @@ annotations etc. As a result, these approaches can struggle in challenging scenarios where there is abrupt cut-in, stopping, overtaking, merging, etc from the neighbouring vehicles. -In this paper, we propose UAP-BEV, a novel approach -that models the noise in Spatio-Temporal BEV predictions -to create an uncertainty-aware occupancy grid map. Using -queries of the distance to the closest occupied cell, we obtain a -sample estimate of the collision probability of the ego-vehicle. -Subsequently, our approach uses gradient-free sampling-based -optimization to compute low-cost trajectories over the cost -map. Importantly, the sampling distribution is adapted based -on the optimal cost values of the sampled trajectories. By -explicitly modelling probabilistic collision avoidance in the -BEV space, our approach is able to outperform the cost-map- -based baselines in collision avoidance, route completion, time -to completion, and smoothness. - -## Scenarios Simulated - - +In this paper, we propose UAP-BEV, a novel approach that models the noise in Spatio-Temporal BEV predictions to create an uncertainty-aware occupancy grid map. Using queries of the distance to the closest occupied cell, we obtain a sample estimate of the collision probability of the ego-vehicle. Subsequently, our approach uses gradient-free sampling-based optimization to compute low-cost trajectories over the cost map. Importantly, the sampling distribution is adapted based on the optimal cost values of the sampled trajectories. By explicitly modelling probabilistic collision avoidance in the BEV space, our approach is able to outperform the cost-map based baselines in collision avoidance, route completion, time to completion, and smoothness. [//]: # (Paste images in this section before the table) ## Sources of Uncertainity - - @@ -47,28 +33,28 @@ to completion, and smoothness.
Image 1
+- Noise in RGB Images + +- Noise in Intrinsics, error in GPS. + +- Noisy BEV annotations + ## Dealing with Uncertainty - Parametric Assumption ![uncertain4](/README/imgs/uncertain4.png) - Assumes shape of underlying distribution - Can lead to conservative behaviour - Does not handle Noisy Annotations -- + ## We propose …. UAP-BEV ![bev](/README/imgs/bev.png) - - - -## Part 1: Augmenting Uncertainty into Distance Estimates +### Part 1: Augmenting Uncertainty into Distance Estimates ![Estimate](/README/imgs/estimate.png) -## Part 2: Efficient Batch Optimization with Constrained Projection +### Part 2: Efficient Batch Optimization with Constrained Projection ![efficient](/README/imgs/efficient.png) - - - ## Results ![result](/README/imgs/results.png) @@ -90,17 +76,6 @@ to completion, and smoothness. -## Sources of Aleatoric Uncertainty -- Noise in RGB Images -- Noise in Intrinsics, error in GPS. -- Noisy BEV annotations - - - - - - -