Faraz Kunwar

Faraz Kunwar

Auto-generated profile, based on publicly available information

Faraz Kunwar

Department of Robotics and Artificial Intelligence, School of Mechanical and Manufacturing Engineering, National University of Sciences and Technology, Islamabad, Pakistan

Research Focus & Metrics

Academic Career Length

21Years (2006-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
20 November 2006 in IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics)
Verified
Journal Article
Closed access
FKFaraz Kunwar
Faraz Kunwar
Beno BenhabibBeno Benhabib
Beno Benhabib
This correspondence presents a novel online trajectory-planning method for the autonomous robotic interception of moving targets in the presence of dynamic obstacles, i.e., position and velocity matching (also referred to as rendezvous). The proposed time-optimal interception method is a hybrid algorithm that augments a novel rendezvous-guidance (RG) technique with the velocity-obstacle approach, for obstacle avoidance, first reported by Fiorini and Shiller. The obstacle-avoidance algorithm itself could not be used in its original form and had to be modified to ensure that the online planned path deviates minimally from the one generated by the RG algorithm. Extensive simulation and experimental analyses, some of which are reported in this correspondence, have clearly demonstrated the tangible time efficiency of the proposed interception method

1 August 2010
Verified
Conference Paper
Closed access
URU. Rasheed
U. Rasheed
MAM. Ahmed
M. Ahmed
SAS. Ali
S. Ali
JAJ. Afridi
J. Afridi
FKF. Kunwar
F. Kunwar
The detection of driving space is the most fundamental step in intelligent vehicle control. This research paper proposes a generic vision based algorithm for identifying driving surfaces in various indoor and outdoor environments. In this paper, instead of relying on a static model for demarcating the boundaries of the driving surfaces, we propose a novel algorithm that provides an adaptive method to detect a drivable surface in any environment. The uniqueness of the proposed algorithm lies in the robustness of the adaptive model that caters for changes in the environment. These changes may be in the form of light composition, off road disturbances, on road static and dynamic objects, shadows and variations in texture for indoor environment. It basically provides a highly dynamic online mechanism for changing the parameters of the Canny Edge Enhancement algorithm. This enables us to accurately determine the starting point and orientation of the driving surface boundary. Subsequently weighted average is used on the candidate edges to optimize the edge detection results. Experiments were carried out on our university's Intelligent Driving System (IDRIS) for outdoor environments and on P3AT for indoor purposes. The experimentation results show that the proposed method can detect the driving surface boundaries in real-time for various different environments.

6 September 2026 in Scientific Reports
Verified
Journal Article
Open access
PDF file
MKMuhammad Shahzad Alam Khan
Muhammad Shahzad Alam Khan
AAAnas Bin Aqeel
Anas Bin Aqeel
Hassan ElahiHassan Elahi
Hassan Elahi
FKFaraz Kunwar
Faraz Kunwar
DHDANISH HUSSAIN
DANISH HUSSAIN
AMAnjum Naeem Malik
Anjum Naeem Malik
Reliable localization is required for autonomous mobile robots when individual sensing streams become noisy, intermittent, or unavailable. This study evaluates a multi-sensor fusion framework that combines LiDAR, monocular vision, GPS, UWB, and IMU data using three strategies: (i) a baseline Extended Kalman Filter (EKF); (ii) a dual-stage sequential EKF that refines LiDAR-Inertial Odometry (LIO) before the final fusion stage; and (iii) a hybrid learning-filtering approach in which modality-specific learned motion and position estimates are incorporated into an EKF. All evaluations were conducted in ROS-Gazebo under nominal operation and controlled sensor-degradation/dropout conditions. Relative to the controller-derived reference trajectory, the standard EKF achieved 0.2235 m RMSE and the dual-stage EKF achieved 0.2029 m RMSE, a descriptive reduction of approximately 9.2% for the reported run. The hybrid learning-EKF achieved 0.212 m RMSE under nominal sensing and 0.384 m RMSE during the tested failure sequence. These results support the evaluated fusion designs under the reported simulation conditions, but they do not establish statistical generalization or universal real-world resilience; independent ground truth, repeated trials, GPS ablation, and physical validation remain necessary.

12 September 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
NANooh Ayub
Nooh Ayub
SKShahbaz Khan
Shahbaz Khan
KKKunwar Faraz Ahmed Khan
Kunwar Faraz Ahmed Khan
MNMuhammad Saqib Nazir
Muhammad Saqib Nazir
Image restoration tasks such as desnowing and dehazing remain challenging due tocomplex degradations that affect both low- and high-frequency components. ExistingCNN-based methods often rely on Layer Normalization, which enforces fixed statisticaldistributions and can oversmooth fine-grained details, limiting the recovery ofcritical high-frequency information. To address this limitation, we propose a unifiedframework, termed NAFXT, which integrates Dynamic Tanh (DyT) normalization withfrequency-aware modules. Unlike conventional normalization, DyT employs learnable,data-dependent scaling that, when combined with standard convolutional operations,preserves local feature variations and avoids oversmoothing, thereby maintaining essentialhigh-frequency details. This enables the generation of detailed feature mapsthat are essential for subsequent frequency processing. Specifically, the NAFXTBlockacts as a preprocessing stage, while the Multi-Branch Dynamic Selective Frequency(MDSF) and Multi-Branch Compact Selective Frequency (MCSF) modules captureglobal frequency information. This synergistic design facilitates effective local andglobal feature extraction. Experimental results demonstrate that the proposed methodachieves superior performance on both desnowing and dehazing benchmarks, reachingup to 41.25 dB PSNR while training time converges faster than Layer Normalization basedcounterparts. However, the framework has not yet been validated on a broaderrange of restoration tasks, which we leave for future work.