Resolve conda issues

Dockerfile

@@ -8,23 +8,25 @@ ARG DONKEYCAR_VERSION=4.3.22 DONKEYCAR_BRANCH=main
 
 # https://github.com/mamba-org/mamba/issues/1403#issuecomment-1024629004
 RUN conda update conda && \
-    mamba update mamba && \
-    pip install ipykernel==6.7.0
+    mamba update mamba
 
 RUN mkdir /opt/local && \
     cd /opt/local && \
     git clone https://github.com/autorope/donkeycar -b $DONKEYCAR_BRANCH && \
     cd donkeycar && \
     mamba env create -f install/envs/ubuntu.yml && \
     eval "$(conda shell.bash hook)"
+
 RUN mamba install -n donkey cudatoolkit=10.1 tensorflow-gpu=2.2.0 nb_conda_kernels -c anaconda -y && \
     chown -R jovyan /home/jovyan
 
-RUN python -m ipykernel install --name=donkey --display-name="Donkey Car ($DONKEYCAR_VERSION-$DONKEYCAR_BRANCH)"
+RUN conda run -n donkey /bin/bash -c "ipython kernel install --name=donkey --display-name=\"Donkey Car ($DONKEYCAR_VERSION-$DONKEYCAR_BRANCH)\""
 RUN chown -R jovyan /opt/local
 
+WORKDIR /opt/local/donkeycar
+RUN conda run -n donkey /bin/bash -c "pip install -e .[pc]"
+
+WORKDIR /home/jovyan
+
 USER $NB_UID
-RUN cd /opt/local/donkeycar && eval "$(conda shell.bash hook)" && conda activate donkey && \
-    pip install -e .[pc]
-#    conda install -c conda-forge --yes --quiet --verbose  cudatoolkit=10.1 && \