본문 바로가기

Papers

★★★ Useful Expressions for Technical Papers ★★★ 1) Predicting the future trajectory of traffic agents has ____________ in autonomous driving.a) played a crucial roleb) been a critical taskc) been a key challenged) been an essential taske) been a major concernf) a challenging taskg) been of great interest 2) One of the key challenges in the task has been something. 3) Somethings can be divided into four categories.Somethings can be categorize.. 더보기
[펌] 인과관계 영어 용어 정리 (1) Since / As / Because 원인 + 결과 as는 뜻이 많기 때문에 since, because 쓰는게 더 뜻이 와 닿음. (2) 원인. Therefore/Thus/Hence/As a result, 결과. 두 문장으로 분리시켜 쓸 수 있는 간단한 형태이지... 그런데 period (.) 대신 semicolon (;)을 활용할 수도 있음. A; therefore/thus/hence, B 마찬가지로 however도 ; however, 이런식으로 쓸 수 있음. 너무 길어지지 않는 선에서 ; 써서 표현하면 있어보일 수 있음..ㅎㅎ; 실제로 토플이나 GRE writing에서 ; 써보니까 좋은 요소라고 함.. 그런데 너무 남발하면 안 좋음. (3) 결과를 나타날 때 thereby와 thereafter.. 더보기
논문에서 자주 쓰이는 단어 증가와 관련된 어휘: Increase, upregulate, enhance, potentiate, facilitate, promote, stimulate, elevate, augment, enrich, aggravate, accelerate, expand, extend, enlarge, rise, alter 감소와 관련된 어휘: Decrease, downregulate, inhibit, prevent, repress, suppress, abolish, nullify, attenuate, knockout, diminish, mitigate, ameliorate, alleviate, restrain, dampen, reduce, lower, eliminate, lessen, shrink, minimize, .. 더보기
[Lange Change] trajectories and maneuvers of surrounding vehicles with panoramic camera arrays, IEEE Trans. on IV, 2016 Summary . Collect vehicle trajectories from surround view images . But they did not open the dataset !!!!!! W.T.F 더보기
[Lane Change] a combined model and learning based framework for interaction aware maneuver prediction, IEEE Trans. on IV, 2016 Summary . Predict future intentions of the vehicles . First predict future intentions ({LLC, RLC, LK}) by using model-based algorithm . Second, predict maneuvers (lateral={LLC, RLC, LK}, longitunial={accelerations}) from the past trajectories and the future intentions, statistically 1. Input . position, acceleration, dist to the next highway junction, type of lane marking, distance to lane end, .. 더보기
[Lane Change] Multi-Modal Trajectory Prediction of Surrounding Vehicles with Maneuver based LSTMs, IEEE IV, 2018 . Summary 1) uses past trajs of vehicles as input 2) outputs possible trajectories w.r.t 6 maneuvers and the corresponding scores 3) Single RNN is used to maneuver decision probability, Encoder-Decoder based RNN is used for traj. prediction 1. Input . (x, y) coordinate of target and its six neighbors 2. Network [Network Structure] . A single RNN is used to produce prob dist over 6 manuevers . En.. 더보기
[Lane Change] Lane change detection based on vehicle trajectory prediction, IEEE Robotics and Automation Letters, 2017 Summary . Estimate driving-intention first, predict trajectory based on intention next, update driving-intention based on trajectory finally. . update procedure reduces false alarm 1. Input . Distance to centerline, lateral velocity, relative distance between target and its neighbors 2. Algorithm [Main algorithm] . driving intention is first estimated. Four classes as follows. [driving intention.. 더보기
[Lane Change] An LSTM netowkr for highway trajectory prediction, IEEE ITSC, 2017 summary 1. consider 9 neighbors and target 2. used NGSIM dataset 3. state vector includes {Position, Speed, Vehicle Tyep, Relative speed, Relative distance, time-to-collision} 1. Input : {Position, Speed, Vehicle Tyep, Relative speed, Relative distance, time-to-collision} 2. Network structure : simple LSTM 3. Results 더보기
[Lane Change] Predicting future lane changes of other highway vehicles using RNN-based deep models, arXiv, 2018 Summary 1. the state vector that represents the current state of a car includes {Position, Heading, Speed, Yaw rate, the number of lanes to the left, the number of lanes to the right} 2. the state vector of the target as well as those of its six neighbors are used for input 3. 4 LSTMs are used and the network structure is determined based on factor model 1. Input . i-th vehicle's state vector in.. 더보기
[CVPR2017] Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network This paper is about super-resolution (SR) !! Really many many SR algorithms have been proposed in the literature. Before the success of convolutional neural network (CNN), sparse-representation based SR algorithms showed the best SR performance. The introduction of CNN into computer vision feild is truely a big impact. CNN has replaced all the existing records not only in high level vision probl.. 더보기

티스토리툴바