Zhihao Cheng

Zhihao Cheng

Auto-generated profile, based on publicly available information

Zhihao Cheng

School of Computer Science, The University of Sydney, Sydney, NSW, Australia

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

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

Representative Publications

View All Publications
29 January 2021 in Analytical Chemistry
Verified
Journal Article
Closed access
YHYan Huang
Yan Huang
Tailin XuTailin Xu
Tailin Xu
YLYong Luo
Yong Luo
CLConghui Liu
Conghui Liu
XGXUAN GAO
XUAN GAO
XZXueji Zhang
Xueji Zhang

9 August 2021 in IEEE Transactions on Neural Networks and Learning Systems
Verified
Journal Article
Closed access
ZCZhihao Cheng
Zhihao Cheng
LLLiu Liu
Liu Liu
Aishan LiuAishan Liu
Aishan Liu
HSHao Sun
Hao Sun
MFMeng Fang
Meng Fang
DTDacheng Tao
Dacheng Tao
Imitation learning from observation (LfO) is more preferable than imitation learning from demonstration (LfD) because of the nonnecessity of expert actions when reconstructing the expert policy from the expert data. However, previous studies imply that the performance of LfO is inferior to LfD by a tremendous gap, which makes it challenging to employ LfO in practice. By contrast, this article proves that LfO is almost equivalent to LfD in the deterministic robot environment, and more generally even in the robot environment with bounded randomness. In the deterministic robot environment, from the perspective of the control theory, we show that the inverse dynamics disagreement between LfO and LfD approaches zero, meaning that LfO is almost equivalent to LfD. To further relax the deterministic constraint and better adapt to the practical environment, we consider bounded randomness in the robot environment and prove that the optimizing targets for both LfD and LfO remain almost the same in the more generalized setting. Extensive experiments for multiple robot tasks are conducted to demonstrate that LfO achieves comparable performance to LfD empirically. In fact, the most common robot systems in reality are the robot environment with bounded randomness (i.e., the environment this article considered). Hence, our findings greatly extend the potential of LfO and suggest that we can safely apply LfO in practice without sacrificing the performance compared to LfD.

29 May 2023 in IEEE Transactions on Neural Networks and Learning Systems
Verified
Journal Article
Closed access
ZCZhihao Cheng
Zhihao Cheng
KZKaining Zhang
Kaining Zhang
LSLi Shen
Li Shen
DTDacheng Tao
Dacheng Tao
Despite remarkable successes in solving various complex decision-making tasks, training an imitation learning (IL) algorithm with deep neural networks (DNNs) suffers from the high-computational burden. In this work, we propose quantum IL (QIL) with a hope to utilize quantum advantage to speed up IL. Concretely, we develop two QIL algorithms: quantum behavioral cloning (Q-BC) and quantum generative adversarial IL (Q-GAIL). Q-BC is trained with a negative log-likelihood (NLL) loss in an offline manner that suits extensive expert data cases, whereas Q-GAIL works in an inverse reinforcement learning (IRL) scheme, which is online, on-policy, and is suitable for limited expert data cases. For both QIL algorithms, we adopt variational quantum circuits (VQCs) in place of DNNs for representing policies, which are modified with data reuploading and scaling parameters to enhance the expressivity. We first encode classical data into quantum states as inputs, then perform VQCs, and finally measure quantum outputs to obtain control signals of agents. Experiment results demonstrate that both Q-BC and Q-GAIL can achieve comparable performance compared to classical counterparts, with the potential of quantum speedup. To our knowledge, we are the first to propose the concept of QIL and conduct pilot studies, which paves the way for the quantum era.

20 June 2023 in IEEE Transactions on Pattern Analysis and Machine Intelligence
Verified
Journal Article
Closed access
ZCZhihao Cheng
Zhihao Cheng
LSLi Shen
Li Shen
MZMiaoxi Zhu
Miaoxi Zhu
JGJiaxian Guo
Jiaxian Guo
MFMeng Fang
Meng Fang
DTDacheng Tao
Dacheng Tao
Existing safe imitation learning (safe IL) methods mainly focus on learning safe policies that are similar to expert ones, but may fail in applications requiring different safety constraints. In this paper, we propose the Lagrangian Generative Adversarial Imitation Learning (LGAIL) algorithm, which can adaptively learn safe policies from a single expert dataset under diverse prescribed safety constraints. To achieve this, we augment GAIL with safety constraints and then relax it as an unconstrained optimization problem by utilizing a Lagrange multiplier. The Lagrange multiplier enables explicit consideration of the safety and is dynamically adjusted to balance the imitation and safety performance during training. Then, we apply a two-stage optimization framework to solve LGAIL: (1) a discriminator is optimized to measure the similarity between the agent-generated data and the expert ones; (2) forward reinforcement learning is employed to improve the similarity while considering safety concerns enabled by a Lagrange multiplier. Furthermore, theoretical analyses on the convergence and safety of LGAIL demonstrate its capability of adaptively learning a safe policy given prescribed safety constraints. At last, extensive experiments in OpenAI Safety Gym conclude the effectiveness of our approach.

16 June 2023
Verified
Conference Paper
Closed access
PDF file
ZCZhihao Cheng
Zhihao Cheng
PWPeng Wang
Peng Wang
In order to study the sliding mode control method of hypersonic aircraft, the six-degree-of-freedom attitude motion model of hypersonic aircraft was established, and the corresponding linearization and decoupling model was obtained by using dynamic inverse method. On this basis, a sliding mode controller based on exponential approach rate is designed, and the controller is reconstructed by introducing the method of instruction parameterization, which greatly reduces the workload of control parameter tuning. In the end, Continuous Action Reinforcement Learning Automata (CARLA) is improved to optimize the new control parameters. Simulation shows that the algorithm can tune out a set of high-quality control parameters in 100 iterations, enabling the control system to quickly and accurately track a given instruction.

1 January 2026 in IEEE Transactions on Transportation Electrification
Verified
Journal Article
Closed access
YGYushan Gu
Yushan Gu
ZCZhihao Cheng
Zhihao Cheng
ZLZheng Liu
Zheng Liu
SFSirui Fan
Sirui Fan
Zhaoke NingZhaoke Ning
Zhaoke Ning
Zhiqiang MaZhiqiang Ma
Zhiqiang Ma