Scale estimation of monocular SLAM using direct acceleration pair measurements

Yathirajam, B and Vaitheeswaran, SM and Ananda, CM (2019) Scale estimation of monocular SLAM using direct acceleration pair measurements. In: ACM International Conference Proceeding Series.

Full text not available from this repository.
Official URL:


Monocular SLAM is increasingly being used to provide navigation solutions for autonomous systems. In this, a visual inertial solution is commonly used to get the scale factor estimate for the monocular SLAM problem where the acceleration data from IMU is integrated to get the absolute velocity and position estimates. Since the accelerometer data may have bias and inaccuracies, it may lead to accumulation of bias and noise errors resulting in drift from the true value positions. This may require an additional sensor to correct the drift. In this paper, the scale factor is estimated using the average acceleration pair data of SLAM and IMU and without any third sensor. For accounting the noise on average accelerations, the Maximum Likelihood estimator is proposed. The scale factor is therefore recovered without any integration of IMU acceleration avoiding any drift errors, improving the estimation of scale factor values. This scale estimation method is included in the ORB-SLAM algorithm in a separate thread for real-time implementation. The output of maximum likelihood estimator is compared with simple estimators namely arithmetic mean, geometric mean and median. The real time formulation developed is validated in experiments using an off the self commercial OptiTrack motion capture system. The present approach gives a robust and less complex estimate of scale factor purely from camera and IMU in the presence of noise on acceleration pairs.

Item Type: Conference or Workshop Item (Paper)
Subjects: AERONAUTICS > Avionics & Aircraft Instrumentation
ENGINEERING > Electronics and Electrical Engineering
MATHEMATICAL AND COMPUTER SCIENCES > Computer Programming and Software
Depositing User: Mrs SK Pratibha
Date Deposited: 30 Mar 2021 09:07
Last Modified: 30 Mar 2021 09:07

Actions (login required)

View Item View Item