TIM (Task Inversion and quality Monitoring) Framework Version 1.0 (first prototype)
The TIM framework represents a novel approach to robotic task planning and execution, specifically designed to handle structured and invertible tasks in dynamic environments. Developed within the INVERSE project, the framework integrates flexible execution control, adaptive planning, and real-time quality monitoring into a coherent and modular architecture.
At the core of the TIM framework lie three tightly integrated components: the Executive System, the Task Planning Module, and the Quality Monitoring Module. These internal subsystems operate in close coordination to ensure that robots can not only carry out complex tasks but also adapt and recover in case of failure or unforeseen events.
The Executive System, inspired by attentional control models, manages the dynamic execution of tasks. It decomposes high-level plans into executable behaviors, regulates the activation and scheduling of concurrent processes, and orchestrates sensorimotor activity. Central to this system are two memory structures: the Long-Term Memory (LTM), which stores task definitions and schemas, and the Working Memory (WM), which maintains the current task hierarchy and execution state. A Behavior-Based System (BBS) handles low-level actions, dynamically prioritizing behaviors based on a combination of top-down goals and bottom-up stimuli. This design allows the robot to reactively shift focus, collaborate with humans, and resolve resource conflicts intelligently during task execution.
The Task Planning Module supports both forward and inverse task generation. When an inverse task is requested, such as disassembling an object or undoing a prior sequence, the planner follows a staged strategy. It first attempts symbolic inversion using traditional planning methods. If this fails, the system can engage learning-based methods to fill gaps or correct errors, and ultimately defer to human guidance through demonstration if needed.
Complementing these systems is the Quality Monitoring Module, which oversees execution fidelity. It continuously evaluates whether task progress aligns with expected outcomes. When deviations are detected, the system can proactively trigger replanning, engage learning mechanisms, or prompt user intervention. This ensures robustness, especially in uncertain or collaborative settings.
In addition to these core components, TIM interfaces with several external systems that supply essential input data. A Scene Understanding Module provides symbolic and quantitative descriptions of the robot’s environment. A Knowledge Base contains information about available actions and robot capabilities. Moreover, TIM supports integration with Self-Learning and Teaching Modules, enabling the robot to expand its skillset autonomously or via user instruction. These external systems are accessed through ROS2-based interfaces, ensuring modularity and interoperability.
The whole architecture is wrapped into a series of dockerized ROS2 packages. Please check the Docker and ROS2 guides for further details.
Before to start, docker must be installed on your OS. For Ubuntu users you may refer to the following procedure:
# Add Docker's official GPG key:
sudo apt-get update
sudo apt-get install ca-certificates curl gnupg
sudo install -m 0755 -d /etc/apt/keyrings
curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /etc/apt/keyrings/docker.gpg
sudo chmod a+r /etc/apt/keyrings/docker.gpg
# Add the repository to Apt sources:
echo \
"deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.gpg] https://download.docker.com/linux/ubuntu \
$(. /etc/os-release && echo "$VERSION_CODENAME") stable" | \
sudo tee /etc/apt/sources.list.d/docker.list > /dev/null
sudo apt-get update
# Install the Docker packages
sudo apt-get install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin
# Add docker user
sudo groupadd docker
sudo usermod -aG docker $USER
newgrp docker
In order to build the package you have to create an image for seed. NOTE: the image may take up to 3Gb of memory on your drive.
./docker_build.sh [image_name]
# example:
./docker_build.sh tim_img
To run TIM a container for the created image must be started, you can used the provided script:
./docker_run.sh [image_name] [container_name]
# example:
./docker_run.sh tim_img tim_cnt
The shell will be automatically attached to the container. Notice that if you close this shell, the container will be closed as well.
If you want to attach a new shell to the previously started container you can use the following script:
./docker_attach.sh [container_name]
# example:
./docker_attach.sh tim_cnt
To launch TIM software, run the following command from within the container:
ros2 launch seed tim.launch.py
This command will run both the SEED and the task_planning nodes. Notice that the robotic sensorimotor primitives specified in this version are defined for the Blocksworld setup at PrismaLab (IIWA robot endowed with WSG50 Gripper).
See references of specific packages
Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Health and Digital Executive Agency (HADEA). Neither the European Union nor HADEA can be held responsible for them. EU -HE Inverse - Grant Agreement 101136067
This project is licensed under the MIT License. You are free to use, modify, and distribute this software with proper attribution. See the LICENSE file for details.