The University of Tokyo · Computer Science
Professor Chia-Ming Chang's research lab specializes in human-centered intelligent systems, with a primary focus on human-vehicle interaction, particularly communication between autonomous vehicles and pedestrians. The lab investigates multimodal interfaces—such as eye-tracking, text displays, and expressive car surfaces—to enhance safety and understanding in urban mobility. Key research directions include perceptual psychology in autonomous driving, user experience design for self-driving technologies, and improving annotation quality in AI training data through spatial and hierarchical labeling methods. The lab emphasizes real-world applicability, combining experimental studies with human factors to shape safer, more intuitive autonomous transportation systems.
Figures are computed from collected data and may differ slightly.
Self-driving technologies have been increasingly developed and tested in recent years (e.g., Volvo's and Google's self-driving cars). However, only a limited number of investigations have so far been conducted into communication between self-driving cars and pedestrians. For example, when a pedestrian is about to cross a street, that pedestrian needs to know the intension of the approaching self-driving car. In the present study, we designed a novel interface known as "Eyes on a Car" to address
There are increasing needs in communication between an autonomous car and a pedestrian. Some conceptual solutions have been proposed to solve this issue, such as using various communication modalities (eyes, smile, text, light and projector) on a car to communicate with pedestrians. However, there is no detailed study in comparing these communication modalities. In this study, we compare five modalities in a pedestrian street-crossing situation via a video experiment. The results show that a tex
Various car manufacturers and researchers have explored the idea of adding eyes to a car as an additional communication modality. A previous study demonstrated that autonomous vehicles’ (AVs) eyes help pedestrians make faster street-crossing decisions. In this study, we examine a more critical question, "can eyes reduce traffic accidents?” To answer this question, we consider a critical street-crossing situation in which a pedestrian is in a hurry to cross the street. If the car is not looking a
Non-expert annotators (who lack sufficient domain knowledge) are often recruited for manual image labeling tasks owing to the lack of expert annotators. In such a case, label quality may be relatively low. We propose leveraging the spatial layout for improving label quality in non-expert image annotation. In the proposed system, an annotator first spatially lays out the incoming images and labels them on an open space, placing related items together. This serves as a working space (spatial organ
Manual image labeling (selecting an appropriate “category” for an image) is very tedious and time consuming especially when selecting labels from a large number of categories. In this study, we propose a hierarchical assignment of labeling tasks where the labelers recursively classify images in a category group into sub category groups, working on a single level at a time. This significantly makes each labeler's task easier, reducing the number of choices from 1,000 to 27 on average. In the user
Communication between an autonomous car and a pedestrian is an important issue that has been widely discussed in the recent years. Many studies and car companies proposed concepts of communication interface on an autonomous car to communicate with a pedestrian. However, there is no detailed study in exploring communication interfaces from a gender perspective. In this study, we explored vehicle-to-pedestrian communication from a gender perspective. We firstly designed three types of communicatio
Perceived journal preference of successful electronic commerce (EC) researchers is an important cue for subsequent researchers in EC academic community. This article contributes to and develops the discussion on the journal preference patterns of researchers in electronic commerce academic community. Some studies of the EC literature have made some contributions from various views. For examples, Bharati and Tarasewich (2002) especially provided a newly EC journal list, based on the global resear
Learning-based methods have achieved great success on single image dehazing in recent years. However, these methods are often subject to performance degradation when domain shifts are confronted. Specifically, haze density gaps exist among the existing datasets, often resulting in poor performance when these methods are tested across datasets. To address this issue, we propose a density-aware data augmentation method (DAMix) that generates synthetic hazy samples according to the haze density lev
This study presents “Speed Labeling”, an image-labeling technique to increase the efficiency of easy binary labeling tasks where an annotator can choose a label instantly. We first conduct a formative study to identify the factors affecting the efficiency of easy image labeling: image layout and image transition. Based on these results, we designed a novel labeling technique using non- stop scrolling. In conventional image labeling, the system moves to the next image only after the user assigns
A DRAM with multiple-refresh-period (MRP) method is one of effective refresh power reduction techniques. To support the MRP method, effective test methods for classifying the refresh period of each DRAM block are needed. In this paper, we propose an effective test method for classifying the refresh periods of DRAM blocks. Also, a programmable built-in self-test (BIST) scheme being able to support the test method is proposed.
Open papers in the app to read, cite, and organize with AI.
