Determine 1: “Interactive Fleet Studying” (IFL) refers to robotic fleets in business and academia that fall again on human teleoperators when vital and regularly be taught from them over time.
In the previous few years we now have seen an thrilling growth in robotics and synthetic intelligence: giant fleets of robots have left the lab and entered the actual world. Waymo, for instance, has over 700 self-driving vehicles working in Phoenix and San Francisco and is at present increasing to Los Angeles. Different industrial deployments of robotic fleets embrace purposes like e-commerce order achievement at Amazon and Ambi Robotics in addition to meals supply at Nuro and Kiwibot.
Industrial and industrial deployments of robotic fleets: bundle supply (high left), meals supply (backside left), e-commerce order achievement at Ambi Robotics (high proper), autonomous taxis at Waymo (backside proper).
These robots use latest advances in deep studying to function autonomously in unstructured environments. By pooling information from all robots within the fleet, the complete fleet can effectively be taught from the expertise of every particular person robotic. Moreover, because of advances in cloud robotics, the fleet can offload information, reminiscence, and computation (e.g., coaching of huge fashions) to the cloud through the Web. This strategy is named “Fleet Studying,” a time period popularized by Elon Musk in 2016 press releases about Tesla Autopilot and utilized in press communications by Toyota Analysis Institute, Wayve AI, and others. A robotic fleet is a contemporary analogue of a fleet of ships, the place the phrase fleet has an etymology tracing again to flēot (‘ship’) and flēotan (‘float’) in Previous English.
Information-driven approaches like fleet studying, nonetheless, face the issue of the “lengthy tail”: the robots inevitably encounter new situations and edge circumstances that aren’t represented within the dataset. Naturally, we are able to’t count on the longer term to be the identical because the previous! How, then, can these robotics firms guarantee enough reliability for his or her providers?
One reply is to fall again on distant people over the Web, who can interactively take management and “tele-operate” the system when the robotic coverage is unreliable throughout activity execution. Teleoperation has a wealthy historical past in robotics: the world’s first robots had been teleoperated throughout WWII to deal with radioactive supplies, and the Telegarden pioneered robotic management over the Web in 1994. With continuous studying, the human teleoperation information from these interventions can iteratively enhance the robotic coverage and cut back the robots’ reliance on their human supervisors over time. Relatively than a discrete soar to full robotic autonomy, this technique affords a steady various that approaches full autonomy over time whereas concurrently enabling reliability in robotic methods right this moment.
The usage of human teleoperation as a fallback mechanism is more and more common in fashionable robotics firms: Waymo calls it “fleet response,” Zoox calls it “TeleGuidance,” and Amazon calls it “continuous studying.” Final yr, a software program platform for distant driving referred to as Phantom Auto was acknowledged by Time Journal as one in every of their High 10 Innovations of 2022. And simply final month, John Deere acquired SparkAI, a startup that develops software program for resolving edge circumstances with people within the loop.
A distant human teleoperator at Phantom Auto, a software program platform for enabling distant driving over the Web.
Regardless of this rising pattern in business, nonetheless, there was comparatively little concentrate on this subject in academia. In consequence, robotics firms have needed to depend on advert hoc options for figuring out when their robots ought to cede management. The closest analogue in academia is interactive imitation studying (IIL), a paradigm by which a robotic intermittently cedes management to a human supervisor and learns from these interventions over time. There have been a lot of IIL algorithms lately for the single-robot, single-human setting together with DAgger and variants akin to HG-DAgger, SafeDAgger, EnsembleDAgger, and ThriftyDAgger; nonetheless, when and change between robotic and human management continues to be an open drawback. That is even much less understood when the notion is generalized to robotic fleets, with a number of robots and a number of human supervisors.
IFL Formalism and Algorithms
To this finish, in a latest paper on the Convention on Robotic Studying we launched the paradigm of Interactive Fleet Studying (IFL), the primary formalism within the literature for interactive studying with a number of robots and a number of people. As we’ve seen that this phenomenon already happens in business, we are able to now use the phrase “interactive fleet studying” as unified terminology for robotic fleet studying that falls again on human management, relatively than hold monitor of the names of each particular person company answer (“fleet response”, “TeleGuidance”, and so forth.). IFL scales up robotic studying with 4 key parts:
- On-demand supervision. Since people can’t successfully monitor the execution of a number of robots directly and are liable to fatigue, the allocation of robots to people in IFL is automated by some allocation coverage $omega$. Supervision is requested “on-demand” by the robots relatively than inserting the burden of steady monitoring on the people.
- Fleet supervision. On-demand supervision allows efficient allocation of restricted human consideration to giant robotic fleets. IFL permits the variety of robots to considerably exceed the variety of people (e.g., by an element of 10:1 or extra).
- Continuous studying. Every robotic within the fleet can be taught from its personal errors in addition to the errors of the opposite robots, permitting the quantity of required human supervision to taper off over time.
- The Web. Due to mature and ever-improving Web expertise, the human supervisors don’t have to be bodily current. Trendy pc networks allow real-time distant teleoperation at huge distances.
Within the Interactive Fleet Studying (IFL) paradigm, M people are allotted to the robots that want probably the most assist in a fleet of N robots (the place N will be a lot bigger than M). The robots share coverage $pi_{theta_t}$ and be taught from human interventions over time.
We assume that the robots share a typical management coverage $pi_{theta_t}$ and that the people share a typical management coverage $pi_H$. We additionally assume that the robots function in impartial environments with equivalent state and motion areas (however not equivalent states). In contrast to a robotic swarm of usually low-cost robots that coordinate to attain a typical goal in a shared surroundings, a robotic fleet concurrently executes a shared coverage in distinct parallel environments (e.g., totally different bins on an meeting line).
The purpose in IFL is to search out an optimum supervisor allocation coverage $omega$, a mapping from $mathbf{s}^t$ (the state of all robots at time t) and the shared coverage $pi_{theta_t}$ to a binary matrix that signifies which human shall be assigned to which robotic at time t. The IFL goal is a novel metric we name the “return on human effort” (ROHE):
[max_{omega in Omega} mathbb{E}_{tau sim p_{omega, theta_0}(tau)} left[frac{M}{N} cdot frac{sum_{t=0}^T bar{r}( mathbf{s}^t, mathbf{a}^t)}{1+sum_{t=0}^T |omega(mathbf{s}^t, pi_{theta_t}, cdot) |^2 _F} right]]
the place the numerator is the whole reward throughout robots and timesteps and the denominator is the whole quantity of human actions throughout robots and timesteps. Intuitively, the ROHE measures the efficiency of the fleet normalized by the whole human supervision required. See the paper for extra of the mathematical particulars.
Utilizing this formalism, we are able to now instantiate and evaluate IFL algorithms (i.e., allocation insurance policies) in a principled means. We suggest a household of IFL algorithms referred to as Fleet-DAgger, the place the coverage studying algorithm is interactive imitation studying and every Fleet-DAgger algorithm is parameterized by a singular precedence perform $hat p: (s, pi_{theta_t}) rightarrow [0, infty)$ that each robot in the fleet uses to assign itself a priority score. Similar to scheduling theory, higher priority robots are more likely to receive human attention. Fleet-DAgger is general enough to model a wide range of IFL algorithms, including IFL adaptations of existing single-robot, single-human IIL algorithms such as EnsembleDAgger and ThriftyDAgger. Note, however, that the IFL formalism isn’t limited to Fleet-DAgger: policy learning could be performed with a reinforcement learning algorithm like PPO, for instance.
IFL Benchmark and Experiments
To determine how to best allocate limited human attention to large robot fleets, we need to be able to empirically evaluate and compare different IFL algorithms. To this end, we introduce the IFL Benchmark, an open-source Python toolkit available on Github to facilitate the development and standardized evaluation of new IFL algorithms. We extend NVIDIA Isaac Gym, a highly optimized software library for end-to-end GPU-accelerated robot learning released in 2021, without which the simulation of hundreds or thousands of learning robots would be computationally intractable. Using the IFL Benchmark, we run large-scale simulation experiments with N = 100 robots, M = 10 algorithmic humans, 5 IFL algorithms, and 3 high-dimensional continuous control environments (Figure 1, left).
We also evaluate IFL algorithms in a real-world image-based block pushing task with N = 4 robot arms and M = 2 remote human teleoperators (Figure 1, right). The 4 arms belong to 2 bimanual ABB YuMi robots operating simultaneously in 2 separate labs about 1 kilometer apart, and remote humans in a third physical location perform teleoperation through a keyboard interface when requested. Each robot pushes a cube toward a unique goal position randomly sampled in the workspace; the goals are programmatically generated in the robots’ overhead image observations and automatically resampled when the previous goals are reached. Physical experiment results suggest trends that are approximately consistent with those observed in the benchmark environments.
Takeaways and Future Directions
To address the gap between the theory and practice of robot fleet learning as well as facilitate future research, we introduce new formalisms, algorithms, and benchmarks for Interactive Fleet Learning. Since IFL does not dictate a specific form or architecture for the shared robot control policy, it can be flexibly synthesized with other promising research directions. For instance, diffusion policies, recently demonstrated to gracefully handle multimodal data, can be used in IFL to allow heterogeneous human supervisor policies. Alternatively, multi-task language-conditioned Transformers like RT-1 and PerAct can be effective “data sponges” that enable the robots in the fleet to perform heterogeneous tasks despite sharing a single policy. The systems aspect of IFL is another compelling research direction: recent developments in cloud and fog robotics enable robot fleets to offload all supervisor allocation, model training, and crowdsourced teleoperation to centralized servers in the cloud with minimal network latency.
While Moravec’s Paradox has so far prevented robotics and embodied AI from fully enjoying the recent spectacular success that Large Language Models (LLMs) like GPT-4 have demonstrated, the “bitter lesson” of LLMs is that supervised learning at unprecedented scale is what ultimately leads to the emergent properties we observe. Since we don’t yet have a supply of robot control data nearly as plentiful as all the text and image data on the Internet, the IFL paradigm offers one path forward for scaling up supervised robot learning and deploying robot fleets reliably in today’s world.
Acknowledgements
This post is based on the paper “Fleet-DAgger: Interactive Robot Fleet Learning with Scalable Human Supervision” presented at the 6th Annual Conference on Robot Learning (CoRL) in December 2022 in Auckland, New Zealand. The research was performed at the AUTOLab at UC Berkeley in affiliation with the Berkeley AI Research (BAIR) Lab and the CITRIS “People and Robots” (CPAR) Initiative. The authors were supported in part by donations from Google, Siemens, Toyota Research Institute, and Autodesk and by equipment grants from PhotoNeo, NVidia, and Intuitive Surgical. Any opinions, findings, and conclusions or recommendations expressed are those of the authors and do not necessarily reflect the views of the sponsors. Thanks to co-authors Lawrence Chen, Satvik Sharma, Karthik Dharmarajan, Brijen Thananjeyan, Pieter Abbeel, and Ken Goldberg for their contributions and helpful feedback on this work.
For more details on interactive fleet learning, see the paper on arXiv, CoRL presentation video on YouTube, open-source codebase on Github, high-level summary on Twitter, and project website.
If you would like to cite this article, please use the following bibtex:
@article{ifl_blog,
title={Interactive Fleet Learning},
author={Hoque, Ryan},
url={https://bair.berkeley.edu/blog/2023/04/06/ifl/},
journal={Berkeley Artificial Intelligence Research Blog},
year={2023}
}
