Maximilian Du

Maximilian Du, Suraj Nair, Dorsa Sadigh et al. · stanford

Enabling robots to learn novel visuomotor skills in a data-efficient manner remains an unsolved problem with myriad challenges. A popular paradigm for tackling this problem is through leveraging large unlabeled datasets that have many behaviors in them and then adapting a policy to a specific task using a small amount of task-specific human supervision (i.e. interventions or demonstrations). However, how best to leverage the narrow task-specific supervision and balance it with offline data remains an open question. Our key insight in this work is that task-specific data not only provides new data for an agent to train on but can also inform the type of prior data the agent should use for learning. Concretely, we propose a simple approach that uses a small amount of downstream expert data to selectively query relevant behaviors from an offline, unlabeled dataset (including many sub-optimal behaviors). The agent is then jointly trained on the expert and queried data. We observe that our method learns to query only the relevant transitions to the task, filtering out sub-optimal or task-irrelevant data. By doing so, it is able to learn more effectively from the mix of task-specific and offline data compared to naively mixing the data or only using the task-specific data. Furthermore, we find that our simple querying approach outperforms more complex goal-conditioned methods by 20% across simulated and real robotic manipulation tasks from images. See https://sites.google.com/view/behaviorretrieval for videos and code.

Maximilian Du, Olivia Y. Lee, Suraj Nair et al. · stanford

Humans are capable of completing a range of challenging manipulation tasks that require reasoning jointly over modalities such as vision, touch, and sound. Moreover, many such tasks are partially-observed; for example, taking a notebook out of a backpack will lead to visual occlusion and require reasoning over the history of audio or tactile information. While robust tactile sensing can be costly to capture on robots, microphones near or on a robot's gripper are a cheap and easy way to acquire audio feedback of contact events, which can be a surprisingly valuable data source for perception in the absence of vision. Motivated by the potential for sound to mitigate visual occlusion, we aim to learn a set of challenging partially-observed manipulation tasks from visual and audio inputs. Our proposed system learns these tasks by combining offline imitation learning from a modest number of tele-operated demonstrations and online finetuning using human provided interventions. In a set of simulated tasks, we find that our system benefits from using audio, and that by using online interventions we are able to improve the success rate of offline imitation learning by ~20%. Finally, we find that our system can complete a set of challenging, partially-observed tasks on a Franka Emika Panda robot, like extracting keys from a bag, with a 70% success rate, 50% higher than a policy that does not use audio.

Maximilian Du, Jennifer Grannen, Shuran Song et al.

The physicality of exercise makes the role of athletic trainers unique. Their physical presence allows them to guide a student through a motion, demonstrate an exercise, and give intuitive feedback. Robot quadrupeds are also embodied agents with robust agility and athleticism. In our work, we investigate whether a robot quadruped can serve as an effective and enjoyable personal trainer device. We focus on a case study of interval training for runners: a repetitive, long-horizon task where precision and consistency are important. To meet this challenge, we propose SNOOPIE, an autonomous robot quadruped pacer capable of running interval training exercises tailored to challenge a user's personal abilities. We conduct a set of user experiments that compare the robot trainer to a wearable trainer device--the Apple Watch--to investigate the benefits of a physical embodiment in exercise-based interactions. We demonstrate 60.6% better adherence to a pace schedule and were 45.9% more consistent across their running speeds with the quadruped trainer. Subjective results also showed that participants strongly preferred training with the robot over wearable devices across many qualitative axes, including its ease of use (+56.7%), enjoyability of the interaction (+60.6%), and helpfulness (+39.1%). Additional videos and visualizations can be found on our website: https://sites.google.com/view/snoopie

Yuejiang Liu, Jubayer Ibn Hamid, Annie Xie et al.

Predicting and executing a sequence of actions without intermediate replanning, known as action chunking, is increasingly used in robot learning from human demonstrations. Yet, its effects on the learned policy remain inconsistent: some studies find it crucial for achieving strong results, while others observe decreased performance. In this paper, we first dissect how action chunking impacts the divergence between a learner and a demonstrator. We find that action chunking allows the learner to better capture the temporal dependencies in demonstrations but at the cost of reduced reactivity to unexpected states. To address this tradeoff, we propose Bidirectional Decoding (BID), a test-time inference algorithm that bridges action chunking with closed-loop adaptation. At each timestep, BID samples multiple candidate predictions and searches for the optimal one based on two criteria: (i) backward coherence, which favors samples that align with previous decisions; (ii) forward contrast, which seeks samples of high likelihood for future plans. By coupling decisions within and across action chunks, BID promotes both long-term consistency and short-term reactivity. Experimental results show that our method boosts the performance of two state-of-the-art generative policies across seven simulation benchmarks and two real-world tasks. Code and videos are available at https://bid-robot.github.io.

Maximilian Du

This paper improves wind power prediction via weather forecast-contextualized Long Short-Term Memory Neural Network (LSTM) models. Initially, only wind power data was fed to a generic LSTM, but this model performed poorly, with erratic and naive behavior observed on even low-variance data sections. To address this issue, weather forecast data was added to better contextualize the power data, and LSTM modifications were made to address specific model shortcomings. These models were tested through both a Normalized Mean Absolute Error and the Naive Ratio (NR), which is a score introduced by this paper to quantify the unwanted presence of naive character in trained models. Results showed an increased accuracy with the addition of weather forecast data on the modified models, as well as a decrease in naive character. Key contributions include making improved LSTM variants, usage of weather forecast data, and the introduction of a new model performance index.

Maximilian Du

This project develops and trains a Recurrent Neural Network (RNN) that monitors sleeping infants from an auxiliary microphone for cases of Sudden Infant Death Syndrome (SIDS), manifested in sudden or gradual respiratory arrest. To minimize invasiveness and maximize economic viability, an electret microphone, and parabolic concentrator, paired with a specially designed and tuned amplifier circuit, was used as a very sensitive audio monitoring device, which fed data to the RNN model. This RNN was trained and operated in the frequency domain, where the respiratory activity is most unique from noise. In both training and operation, a Fourier transform and an autoencoder compression were applied to the raw audio, and this transformed audio data was fed into the model in 1/8 second time steps. In operation, this model flagged each perceived breath, and the time between breaths was analyzed through a statistical T-test for slope, which detected dangerous trends. The entire model achieved 92.5% accuracy on continuous data and had an 11.25-second response rate on data that emulated total respiratory arrest. Because of the compatibility of the trained model with many off-the-shelf devices like Android phones and Raspberry Pi's, free-standing processing hardware deployment is a very feasible future goal.