Research 研究活動

The Image Processing and Robotics Lab team primarily focuses on an extensive array of studies in audio/video processing, robotics, and their applications to provide technological support to individuals requiring assistance, such as the elderly, people with disabilities, natural disaster victims, and more.

Main keywords: Audio/video Processing, UAV, Artificial Intelligence (AI), Intelligent Assistant Systems, ITS, Aerial/mobile Robotics



Main Research Projects 主な研究プロジェクト

A vision-based method detects the landing spot, and the drones move towards the designated location for a safe landing. Vertical flight control is achieved using PID control, while horizontal flight control is accomplished through PD control.

(A video of so far results →)




Malik Research Compact Video.mp4

A UAV equipped with a loudspeaker and a microphone hovers over the disaster site and broadcasts an audio request for a response from anyone below. The microphone detects the voice of anyone who responds, thereby determining whether there is anyone who requires rescuing. However, the microphone also picks up the sound of the nearby UAV propellers, which can obscure the person's voice. This is one of the major challenges that needs to be addressed.




Wireless communication is essential for remote control of these robots, yet using radio waves is unfavorable in the medical domain due to its potential impact on patients and medical equipment. Hence, we present a teleoperation system for a wheeled mobile robot employing visible light, which uses a camera and an array of LEDs. This approach ensures minimal disruption to both patients and medical equipment. (A video of so far results →)



Passenger(add LED Array text).mp4

Tactile paving detection is performed through color image analysis. (A video of so far results →)



Compack video.mp4

The detection of all moving objects at an intersection is achieved by employing a Gaussian mixture model-based approach for moving object detection. The tracking of these moving objects is conducted using the Kalman filter technique.

(A video of so far results →)



(Compact Video)Vehicle detection from Omnidorectional camera.mp4

Omnidirectional camera images typically exhibit lower resolution compared to conventional cameras, making it challenging to discern objects located at a distance from the camera. To tackle this issue, we introduce a hybrid camera system. This system initially identifies less distinct target regions within the omnidirectional view and subsequently employs a pan-tilt camera to capture high-resolution images of these targets.

(A video of  hybrid camera system operation →)




Object capturing near object by hybrid camera system.mp4