Fanbo Zhao

SmartFly: from perception to action

Sun, 30 Aug 2026 00:00:00 +0000

Research in progress · August 2026 snapshot

SmartFly asks: how can an agent use vision, odor, and wind to stay stable and find a target? Built around a MuJoCo fly simulation, the project combines a learned high-level navigation policy with a frozen low-level flight controller to explore the relationship between perception, memory, and action.

Research questions

Can one policy support two navigation tasks? The visual task provides relative target information. The vision–odor task hides the source location and uses odor, apparent wind, and vision for source seeking and obstacle avoidance. The goal is to support both tasks with the same checkpoint.
How do visual priors relate to representations and behavior? The comparison uses a trainable CNN and FlyVis with a frozen pretrained backbone, under matched tasks and training budgets.
What separates a demonstration from reliable evidence? Stable-flight replays help inspect closed-loop behavior and actual policy inputs. Full navigation claims require independent testing and statistics across training seeds.

Current implementation

The project uses Python, PyTorch, Ray RLlib / PPO, MuJoCo, and dm_control, with four main components:

Fixing an AnyDesk connection on Ubuntu 22.04

Sat, 22 Nov 2025 00:00:00 +0000

The issue

When connecting remotely to an Ubuntu 22.04 computer using AnyDesk, the connection failed with this message:

The session has ended.
Status: display_server_not_supported

Some security-related settings on the Ubuntu computer also could not be unlocked.

Investigating the display server

Ubuntu may be running Wayland instead of Xorg. In this setup, switching to Xorg resolved the AnyDesk connection issue.

Check the current session type:

echo $XDG_SESSION_TYPE

Switch to Xorg for the session

Log out of Ubuntu without shutting down the computer.
On the login screen, choose the gear icon and select Ubuntu on Xorg.
Log back in and try the connection again.

Make Xorg the default, if needed

Edit the GDM configuration:

A practical Conda command reference

Mon, 10 Nov 2025 00:00:00 +0000

Create an environment

# Create an environment with a specific Python version
conda create -n myenv python=3.10

# Install packages while creating the environment
conda create -n myenv python=3.10 numpy pandas

Activate and deactivate

conda activate myenv
conda deactivate

List and remove environments

conda env list

# Remove the environment and all packages inside it
conda remove -n myenv --all

Install and remove packages

conda install numpy
conda install numpy=1.26
conda remove numpy

Update packages

conda update numpy
conda update conda
conda update --all

Export and recreate an environment

# Export the environment, including channel information
conda env export > environment.yml

# Recreate the environment from the exported file
conda env create -f environment.yml

Clone an environment and inspect packages

conda create --name newenv --clone oldenv
conda list

About me

Mon, 01 Jan 0001 00:00:00 +0000

Hi, I’m Fanbo Zhao, a second-year undergraduate in artificial intelligence at Fudan University.

My interests include embodied intelligence, reinforcement learning, multimodal perception, and biologically inspired vision. I’m particularly interested in how sensory information helps an agent act in its environment.

My current project exploration centers on SmartFly: fly-inspired navigation in simulation, combining visual, odor, and wind information, with a planned comparison of CNN and FlyVis visual encoders. This is ongoing work; the project page distinguishes implemented components and stabilization replays from research questions still awaiting evaluation.

Get in touch

Mon, 01 Jan 0001 00:00:00 +0000

Have a research or internship opportunity in mind, or simply want to talk about AI? I’d love to hear from you.

zhaofanbo618@gmail.com

GitHub · FanboZhao